Weapon of choice? Musk’s DOGE used Meta’s Llama 2—not Grok—for gov’t slashing, report says Grok apparently wasn't an option. Ashley Belanger – May 22, 2025 5:12 pm | 19 Credit: Anadolu / Contributor | Anadolu Credit: Anadolu / Contributor | Anadolu Story text Size Small Standard Large Width * Standard Wide Links Standard Orange * Subscribers only   Learn more An outdated Meta AI model was apparently at the center of the Department of Government Efficiency's initial ploy to purge parts of the federal government. Wired reviewed materials showing that affiliates of Elon Musk's DOGE working in the Office of Personnel Management "tested and used Meta’s Llama 2 model to review and classify responses from federal workers to the infamous 'Fork in the Road' email that was sent across the government in late January." The "Fork in the Road" memo seemed to copy a memo that Musk sent to Twitter employees, giving federal workers the choice to be "loyal"—and accept the government's return-to-office policy—or else resign. At the time, it was rumored that DOGE was feeding government employee data into AI, and Wired confirmed that records indicate Llama 2 was used to sort through responses and see how many employees had resigned. Llama 2 is perhaps best known for being part of another scandal. In November, Chinese researchers used Llama 2 as the foundation for an AI model used by the Chinese military, Reuters reported. Responding to the backlash, Meta told Reuters that the researchers' reliance on a “single" and "outdated" was "unauthorized," then promptly reversed policies banning military uses and opened up its AI models for US national security applications, TechCrunch reported. "We are pleased to confirm that we’re making Llama available to US government agencies, including those that are working on defense and national security applications, and private sector partners supporting their work," a Meta blog said. "We’re partnering with companies including Accenture, Amazon Web Services, Anduril, Booz Allen, Databricks, Deloitte, IBM, Leidos, Lockheed Martin, Microsoft, Oracle, Palantir, Scale AI, and Snowflake to bring Llama to government agencies." Because Meta's models are open-source, they "can easily be used by the government to support Musk’s goals without the company’s explicit consent," Wired suggested. It's hard to track where Meta's models may have been deployed in government so far, and it's unclear why DOGE relied on Llama 2 when Meta has made advancements with Llama 3 and 4. Not much is known about DOGE's use of Llama 2. Wired's review of records showed that DOGE deployed the model locally, "meaning it’s unlikely to have sent data over the Internet," which was a privacy concern that many government workers expressed. In an April letter sent to Russell Vought, director of the Office of Management and Budget, more than 40 lawmakers demanded a probe into DOGE's AI use, which, they warned—alongside "serious security risks"—could "have the potential to undermine successful and appropriate AI adoption." That letter called out a DOGE staffer and former SpaceX employee who supposedly used Musk’s xAI Grok-2 model to create an "AI assistant," as well as the use of a chatbot named "GSAi"—"based on Anthropic and Meta models"—to analyze contract and procurement data. DOGE has also been linked to a software called AutoRIF that supercharges mass firings across the government. In particular, the letter emphasized the "major concerns about security" swirling DOGE's use of "AI systems to analyze emails from a large portion of the two million person federal workforce describing their previous week’s accomplishments," which they said lacked transparency. Those emails came weeks after the "Fork in the Road" emails, Wired noted, asking workers to outline weekly accomplishments in five bullet points. Workers fretted over responses, worried that DOGE might be asking for sensitive information without security clearances, Wired reported. Wired could not confirm if Llama 2 was also used to parse these email responses, but federal workers told Wired that if DOGE was "smart," then they'd likely "reuse their code" from the "Fork in the Road" email experiment. Why didn’t DOGE use Grok? It seems that Grok, Musk's AI model, wasn't available for DOGE's task because it was only available as a proprietary model in January. Moving forward, DOGE may rely more frequently on Grok, Wired reported, as Microsoft announced it would start hosting xAI’s Grok 3 models in its Azure AI Foundry this week, The Verge reported, which opens the models up for more uses. In their letter, lawmakers urged Vought to investigate Musk's conflicts of interest, while warning of potential data breaches and declaring that AI, as DOGE had used it, was not ready for government. "Without proper protections, feeding sensitive data into an AI system puts it into the possession of a system’s operator—a massive breach of public and employee trust and an increase in cybersecurity risks surrounding that data," lawmakers argued. "Generative AI models also frequently make errors and show significant biases—the technology simply is not ready for use in high-risk decision-making without proper vetting, transparency, oversight, and guardrails in place." Although Wired's report seems to confirm that DOGE did not send sensitive data from the "Fork in the Road" emails to an external source, lawmakers want much more vetting of AI systems to deter "the risk of sharing personally identifiable or otherwise sensitive information with the AI model deployers." A seeming fear is that Musk may start using his own models more, benefiting from government data his competitors cannot access, while potentially putting that data at risk of a breach. They're hoping that DOGE will be forced to unplug all its AI systems, but Vought seems more aligned with DOGE, writing in his AI guidance for federal use that "agencies must remove barriers to innovation and provide the best value for the taxpayer." "While we support the federal government integrating new, approved AI technologies that can improve efficiency or efficacy, we cannot sacrifice security, privacy, and appropriate use standards when interacting with federal data," their letter said. "We also cannot condone use of AI systems, often known for hallucinations and bias, in decisions regarding termination of federal employment or federal funding without sufficient transparency and oversight of those models—the risk of losing talent and critical research because of flawed technology or flawed uses of such technology is simply too high." Ashley Belanger Senior Policy Reporter Ashley Belanger Senior Policy Reporter Ashley is a senior policy reporter for Ars Technica, dedicated to tracking social impacts of emerging policies and new technologies. She is a Chicago-based journalist with 20 years of experience. 19 Comments

The rapid rise of AI agents is sparking both excitement and alarm. Their power lies in their ability to complete tasks with increasing autonomy. Many can already pursue multi-step goals, make decisions, and interact with external systems — all with minimal human input. Teams of AI agents are beginning to collaborate, each handling a specialised role. As their autonomy increases, they’re poised to reshape countless business processes. Tech giants are heralding them as the future of the web. At Microsoft’s Build conference this week, the company declared that we have entered “the era of AI agents.” OpenAI CEO Sam Altman joined the event, proclaiming his lab’s new Codex tool as “a real agentic coding experience.” He called it “one of the biggest changes to programming that I’ve ever seen.” Beyond the hype, practical applications are rapidly emerging. AI agents are already assisting with various tasks, from code generation and cyber threat detection to customer service enquiries and shopping marketing campaigns.  Grab that deal Before long, they could become comprehensive executive assistants — managing your emails, calendar, and projects. But to harness the opportunities, people need to prepare now. Cihangir Kocak is helping them do just that. A principal business and AI strategist at HP, Kocak guides organisations through digital transformation. He believes AI agents will unleash a new wave of opportunities. “We are going to a future where everyone will have an AI agent as an assistant,” he says. At TNW Conference this summer, Kocak will host two sessions on AI agents. On June 19, he’ll deliver a keynote on their rise. The next day, he’ll join Joost Bos, Senior AI Engineer at Deloitte, for a masterclass titled “Agentic AI: Architecting the Future of Business.” Ahead of the event, he shared a few of his tips. 1. Understand what AI agents can do AI agents evolve large language models (LLMs) from passive responders into active problem-solvers. With tools, memory, and defined goals, they can complete complex tasks on their own. “Large language models act as the brains and AI agents as the hands, which means they can also act,” Kocak says. “They can do things for you autonomously.” Agents can also collaborate. One might source products, another handle logistics, a third build your website, and a fourth write the marketing copy. In future, businesses may need their own agents to interact with others. Your AI assistant could collaborate with them to book the best service for your needs. Free courses from the likes of Hugging Face, Salesforce, and Microsoft are good starting points to explore the possibilities. After getting an understanding of the basics, you can put them into practice.  2. Start experimenting Kocak expects AI agents to rapidly reshape workplaces. “I believe that within five years, everything will be changed because of AI agents,” he says. “It might be even much less than five years — maybe two to three years.” Many companies are already shifting numerous tasks from humans to AI. In the near future, the people that they do recruit may require experience of working with AI agents. “Soon, a lot of these companies will ask for people who can work with AI agents,” says Kocak. His advice? “Get your hands dirty. Play with it, experiment with it — but do it consciously.” One tool he recommends is LM Studio, a desktop app for running LLMs locally. But his key recommendation is simply getting started. “Just do something to get a feel of it. Once you have that, it’s time for the next step.” 3. Find use cases After testing some tools, Kocak suggests identifying where they can add value. He advises looking for tasks where AI can free up your time — and start small. “What costs you the most time? What don’t you like to do? When you figure out those things, you can look at how AI agents can help you.” Kocak uses local LLMs for privacy-sensitive tasks, and ChatGPT for public ones — like drafting LinkedIn posts in his own voice. “It saves at least half of my time,” he says. 4. Focus on the data The real magic of AI agents emerges when they’re personalised with your choice of data. Generic tools like ChatGPT can handle broad tasks. But if you want something tailored, agents trained on your choice of data can offer sharper performance. That internal knowledge can turn a generic agent into a bespoke powerhouse. “What makes an AI solution special is when you feed it with your own data,” says Kocak. “Then you will have a solution that can operate differently than anything else.” 5. Maintain human oversight Although AI agents can act autonomously, human oversight remains vital. Agents are powerful, but not flawless. Giving them too much freedom is risky. “It’s wise to have a human in the room,” he says. “The future will be AI agents plus humans — that will be the most beneficial combination.” 6. Stay secure As AI tools become more accessible, security concerns are mounting. Among the threats are data leaks, adversarial attacks, and agents going off the rails. There’s also the risk of losing a competitive edge.  “External parties can take your data and send it to their servers,” says Kocak. “They can then use all sensitive data in your conversations to optimise their models.” Many risks can be reduced by deploying open-source, local models — especially for sensitive data and use cases. “If you really want a competitive advantage, you need to run and own your AI. That sets you apart,” says Kocak. He adds that people shouldn’t be fearful, but conscious. Closed-source, cloud-based tools such as ChatGPT remain useful — but sensitive data and tasks may require more secure alternatives. “Just be aware of what information you enter. And remember there is another, better option, of running your large language model locally.” 7. Embrace the future As the industrial revolution and factory automation did before them, AI agents will transform jobs. Some roles will disappear — but new ones will emerge. A welder could become an operator of robotic welders. A data entry clerk might oversee AI agents. Kocak is optimistic about the possibilities. “Our core capabilities as humans — like being creative, finding solutions out of the box, and empathy — will come to the forefront.” These tips are just a glimpse of what Kocak will provide at TNW Conference. If you want to check out his sessions — or anything else on the event agenda — we have a special offer for you. Use the code TNWXMEDIA2025 at the ticket checkout to get 30% off. Story by Thomas Macaulay Managing editor Thomas is the managing editor of TNW. He leads our coverage of European tech and oversees our talented team of writers. Away from work, he e (show all) Thomas is the managing editor of TNW. He leads our coverage of European tech and oversees our talented team of writers. Away from work, he enjoys playing chess (badly) and the guitar (even worse). Get the TNW newsletter Get the most important tech news in your inbox each week. Also tagged with

The rapid rise of AI agents is sparking both excitement and alarm. Their power lies in their ability to complete tasks with increasing autonomy. Many can already pursue multi-step goals, make decisions, and interact with external systems — all with minimal human input. Teams of AI agents are beginning to collaborate, each handling a specialised role. As their autonomy increases, they’re poised to reshape countless business processes. Tech giants are heralding them as the future of the web. At Microsoft’s Build conference this week, the company declared that we have entered “the era of AI agents.” OpenAI CEO Sam Altman joined the event, proclaiming his lab’s new Codex tool as “a real agentic coding experience.” He called it “one of the biggest changes to programming that I’ve ever seen.” Beyond the hype, practical applications are rapidly emerging. AI agents are already assisting with various tasks, from code generation and cyber threat detection to customer service enquiries and shopping marketing campaigns.  Before long, they could become comprehensive executive assistants — managing your emails, calendar, and projects. But to harness the opportunities, people need to prepare now. Cihangir Kocak is helping them do just that. A principal business and AI strategist at HP, Kocak guides organisations through digital transformation. He believes AI agents will unleash a new wave of opportunities. “We are going to a future where everyone will have an AI agent as an assistant,” he says. At TNW Conference this summer, Kocak will host two sessions on AI agents. On June 19, he’ll deliver a keynote on their rise. The next day, he’ll join Joost Bos, Senior AI Engineer at Deloitte, for a masterclass titled “Agentic AI: Architecting the Future of Business.” Ahead of the event, he shared a few of his tips. 1. Understand what AI agents can do AI agents evolve large language models (LLMs) from passive responders into active problem-solvers. With tools, memory, and defined goals, they can complete complex tasks on their own. “Large language models act as the brains and AI agents as the hands, which means they can also act,” Kocak says. “They can do things for you autonomously.” Agents can also collaborate. One might source products, another handle logistics, a third build your website, and a fourth write the marketing copy. In future, businesses may need their own agents to interact with others. Your AI assistant could collaborate with them to book the best service for your needs. Free courses from the likes of Hugging Face, Salesforce, and Microsoft are good starting points to explore the possibilities. After getting an understanding of the basics, you can put them into practice.  2. Start experimenting Kocak expects AI agents to rapidly reshape workplaces. “I believe that within five years, everything will be changed because of AI agents,” he says. “It might be even much less than five years — maybe two to three years.” Many companies are already shifting numerous tasks from humans to AI. In the near future, the people that they do recruit may require experience of working with AI agents. “Soon, a lot of these companies will ask for people who can work with AI agents,” says Kocak. His advice? “Get your hands dirty. Play with it, experiment with it — but do it consciously.” One tool he recommends is LM Studio, a desktop app for running LLMs locally. But his key recommendation is simply getting started. “Just do something to get a feel of it. Once you have that, it’s time for the next step.” 3. Find use cases After testing some tools, Kocak suggests identifying where they can add value. He advises looking for tasks where AI can free up your time — and start small. “What costs you the most time? What don’t you like to do? When you figure out those things, you can look at how AI agents can help you.” Kocak uses local LLMs for privacy-sensitive tasks, and ChatGPT for public ones — like drafting LinkedIn posts in his own voice. “It saves at least half of my time,” he says. 4. Focus on the data The real magic of AI agents emerges when they’re personalised with your choice of data. Generic tools like ChatGPT can handle broad tasks. But if you want something tailored, agents trained on your choice of data can offer sharper performance. That internal knowledge can turn a generic agent into a bespoke powerhouse. “What makes an AI solution special is when you feed it with your own data,” says Kocak. “Then you will have a solution that can operate differently than anything else.” 5. Maintain human oversight Although AI agents can act autonomously, human oversight remains vital. Agents are powerful, but not flawless. Giving them too much freedom is risky. “It’s wise to have a human in the room,” he says. “The future will be AI agents plus humans — that will be the most beneficial combination.” 6. Stay secure As AI tools become more accessible, security concerns are mounting. Among the threats are data leaks, adversarial attacks, and agents going off the rails. There’s also the risk of losing a competitive edge.  “External parties can take your data and send it to their servers,” says Kocak. “They can then use all sensitive data in your conversations to optimise their models.” Many risks can be reduced by deploying open-source, local models — especially for sensitive data and use cases. “If you really want a competitive advantage, you need to run and own your AI. That sets you apart,” says Kocak. He adds that people shouldn’t be fearful, but conscious. Closed-source, cloud-based tools such as ChatGPT remain useful — but sensitive data and tasks may require more secure alternatives. “Just be aware of what information you enter. And remember there is another, better option, of running your large language model locally.” 7. Embrace the future As the industrial revolution and factory automation did before them, AI agents will transform jobs. Some roles will disappear — but new ones will emerge. A welder could become an operator of robotic welders. A data entry clerk might oversee AI agents. Kocak is optimistic about the possibilities. “Our core capabilities as humans — like being creative, finding solutions out of the box, and empathy — will come to the forefront.” These tips are just a glimpse of what Kocak will provide at TNW Conference. If you want to check out his sessions — or anything else on the event agenda — we have a special offer for you. Use the code TNWXMEDIA2025 at the ticket checkout to get 30% off. Story by Thomas Macaulay Managing editor Thomas is the managing editor of TNW. He leads our coverage of European tech and oversees our talented team of writers. Away from work, he e (show all) Thomas is the managing editor of TNW. He leads our coverage of European tech and oversees our talented team of writers. Away from work, he enjoys playing chess (badly) and the guitar (even worse). Get the TNW newsletter Get the most important tech news in your inbox each week. Also tagged with

Wells Fargo says in a report that President Donald Trump's tariffs won't bring manufacturing back. Carlos Barria/REUTERS 2025-05-23T02:31:46Z Save Saved Read in app This story is available exclusively to Business Insider subscribers. Become an Insider and start reading now. Have an account? Wells Fargo said in a report that President Donald Trump's tariffs won't bring manufacturing back. High labor costs and a lack of workers would make building more factories an "uphill battle." US manufacturing needs $2.9 trillion in investment to reach 1979 employment levels. President Donald Trump's push to revive American manufacturing through tariffs may face some hurdles.Despite some high-profile commitments, including Nvidia's plans for a US-based supercomputer plant and Apple's pledge to invest $500 billion domestically, a new report from Wells Fargo economists predicts that bringing back offshored manufacturing jobs will be an "uphill battle.""An aim of tariffs is to spur a durable rebound in US manufacturing employment," Wells Fargo analysts wrote in the report. "However, a meaningful increase in factory jobs does not appear likely in the foreseeable future, in our view."The report attributes the potentially low factory job growth to high labor costs, a lack of suitable workers to fill vacant positions, and a subdued population growth from lower fertility rates and slower immigration."Higher prices and policy uncertainty may weigh on firms' ability and willingness to expand payrolls," the analysts added.The tariffs are part of Trump's broader economic agenda to revive American manufacturing as a pathway toward middle-class prosperity. The tariffs are meant to hike the costs of imports to incentivize companies to make goods domestically."Jobs and factories will come roaring back into our country," Trump said while announcing tariffs on April 2. "And ultimately, more production at home will mean stronger competition and lower prices for consumers."Some tariffs imposed on April 2 have been temporarily paused or greatly reduced, including tariffs on China. The 10% across-the-board tariff remains, as do some specific tariffs on Mexico and Canada, plus 30% in duties on China. Duties at their current level are still the highest they have been since the 1940s."In order for manufacturing employment to return to its historic peak, we estimate at a minimum $2.9 trillion in net new capital investment is required," Wells Fargo analysts wrote. "Assuming businesses are willing and able to invest such ample sums, questions over staffing remain."The Wall Street bank says that US manufacturing employment currently stands at 12.8 million, down from its 1979 peak of 19.5 million. To get back to that mark, the US would need to add roughly 6.7 million jobs. Wells Fargo added that the figure is nearly the same as the entire pool of unemployed Americans, which in April was 7.2 million, according to the US Bureau of Labor Statistics."Population aging, negative perceptions, and skill mismatches also underpin workforce concerns," Wells Fargo analysts wrote. "New jobs will require different skills than those previously lost."In 2024, Taiwanese chipmaker TSMC said it delayed the opening of its Arizona chip factory due to a shortage of skilled workers. A report released in April 2024 by Deloitte and the Manufacturing Institute also found that nearly half of the 3.8 million new manufacturing jobs anticipated by 2033 could remain unfilled due to skill gaps and other population factors."Tariffs must be high enough to make the cost of domestic production competitive in the US market, and they also must be kept in place long enough for producers to bring on additional workers and expand capacity," the report concluded. "If the economic or political costs are deemed too high, the current administration could quickly dial-back prevailing duties further."The White House did not immediately respond to a request for comments.

In February 2024, CNN reported, “A finance worker at a multinational firm was tricked into paying out $25 million to fraudsters using deepfake technology to pose as the company’s chief financial officer in a video conference call.” In Europe, a second firm experienced a multimillion-dollar fraud when a deepfake emulated a board member in a video allegedly approving a fraudulent transfer of funds. “Banks and financial institutions are particularly at risk,” said The Hack Academy. “A study by Deloitte found that over 50% of senior executives expect deepfake scams to target their organizations soon. These attacks can undermine trust and lead to significant financial loss.”  Hack Academy went on to say that AI-inspired security attacks weren’t confined to deepfakes. These attacks were also beginning to occur with increased regularity in the form of corporate espionage and misinformation campaigns. AI brings new, more dangerous tactics to traditional security attack methods like phishing, social engineering and the insertion of malware into systems. For CIOs, enterprise AI system developers, data scientists and IT network professionals, AI changes the rules and the tactics for security, given AI’s limitless potential for both good and bad. This is forcing a reset in how IT thinks about security against malicious actors and intruders. Related:How Bad Actors are Exploiting AI What exactly is IT up against? The AI tools that are available on the dark web and in public cyber marketplaces give security perpetrators a wide choice of AI weaponry. Also, IoT and edge networks now present much broader enterprise attack surfaces. Security threats can come in videos, phone calls, social media sites, corporate systems and networks, vendor clouds, IoT devices, network end points, and virtually any entry point into a corporate IT environment that electronic communications can penetrate. Here are some of the current AI-embellished security attacks that companies are seeing: Convincing deepfake videos of corporate executives and stakeholders that are intended to dupe companies in pursuing certain actions or transferring certain assets or funds. This deep faking also extends to voice simulations of key personnel that are left as voicemails in corporate phone systems.  Phishing and spearfishing attacks that send convincing emails (some with malicious attachments) to employees, who mistakenly open them because they think the sender is their boss, the CEO or someone else they perceive as trusted. AI supercharges these attacks because it can automate and send out a large volume of emails that hit many employee email accounts. That AI continues to “learn” with the help of machine learning so it can discover new trusted sender candidates for future attacks.   Related:Adaptive messaging that uses generative AI to craft messages to users that correct grammar and that “learn” from corporate communication styles so they can more closely emulate corporate communications that make them seem legitimate. Mutating code that uses AI to change malware signatures on the fly so antivirus detection mechanisms can be evaded. Data poisoning that occurs when a corporate or cloud provider’s AI data repository is injected by malware that alters (“poisons”) so the data produces erroneous and misleading results.  Fighting Back With Tech To combat these supercharged AI-based security threats, IT has number of tools, techniques and strategies it can consider. Fighting deepfakes. Deepfakes can come in the form of videos, voicemails and photos. Since deepfakes are unstructured data objects that can’t be parsed in their native forms like real data, there are new tools on the market that can convert these objects into graphical representations that can be analyzed to evaluate whether there is something in an object that should or shouldn’t be there. The goal is to confirm authenticity.  Related:Fighting phishing and spear phishing. A combination of policy and practice works best to combat phishing and spear phishing attacks. Both types of attacks are predicated on users being tricked into opening an email attachment that they believe is from a trusted sender, so the first line of defense is educating (and repeat-educating) users on how to handle their email. For instance, a user should notify IT if they receive an email that seems unusual or unexpected, and they should never open it. IT should also review its current security tools. Is it still using older security monitoring software that doesn’t include more modern technologies like observability, which can check for security intrusions or malware at more atomic levels?  Is IT still using IAM (identity access management) software to track user identities and activities at a top level in the cloud and on top and atomic levels on premises, or has it also added cloud identity entitlements management (CIEM), which gives it an atomic level view of  user accesses and activities in the cloud? Better yet, has IT moved to identity governance administration (IGA), which can serve as an over-arching umbrella for IAM and CIEM plugins, plus provide detailed audit reports and automated compliance across all platforms? Fighting embedded malware code. Malware can lie dormant in systems for months, giving a bad actor the option to activate it whenever the timing is right. It’s all the more reason for IT to augment its security staff with new skillsets, such as that of the “threat hunter,” whose job is to examine networks, data and systems on a daily basis, hunting down malware that might be lurking within, and destroying it before it activates. Fighting with zero-trust networks. Internet of Things (IoT) devices come into companies with little or no security because IoT suppliers don’t pay much attention to it and there is a general expectation that corporate IT will configure devices to the appropriate security settings. The problem is, IT often forgets to do this. There are also times when users purchase their own IoT gear, and IT doesn’t know about it. Zero-trust networks help manage this, because they detect and report on everything that is added, subtracted or modified on the network. This gives IT visibility into new, potential security breach points. A second step is to formalize IT procedures for IoT devices so that no IoT device is deployed without the device’s security first being set to corporate standards.  Fighting AI data poisoning. AI models, systems and data should be continuously monitored for accuracy. As soon as they show lowered levels of accuracy or produce unusual conclusions, the data repository, inflows and outflows should be examined for quality and non-bias of data. If contamination is found, the system should be taken down, the data sanitized, and the sources of the contamination traced, tracked and disabled. Fighting AI with AI. Most every security tool on the market today contains AI functionality to detect anomalies, abnormal data patterns and user activities. Additionally, forensics AI can dissect a security breach that does occur, isolating how it happened, where it originated from and what caused it. Since most sites don’t have on-staff forensics experts, IT will have to train staff in forensics skills. Fighting with regular audits and vulnerability testing. Minimally, IT vulnerability testing should be performed on a quarterly basis, and full security audits on an annual basis. If sites use cloud providers, they should request each provider’s latest security audit for review. An outside auditor can also help sites prepare for future AI-driven security threats, because auditors stay on top of the industry, visit many different companies, and see many different situations. An advanced knowledge of threats that loom in the future helps sites prepare for new battles. Summary AI technology is moving faster than legal rulings and regulations. This leaves most IT departments “on their own” to develop security defenses against bad actors who use AI against them.  The good news is that IT already has insights into how bad actors intend to use AI, and there are tools on the market that can help defensive efforts. What’s been missing is a proactive and aggressive battle plan from IT. That has to start now. 

In February 2024, CNN reported, “A finance worker at a multinational firm was tricked into paying out $25 million to fraudsters using deepfake technology to pose as the company’s chief financial officer in a video conference call.” In Europe, a second firm experienced a multimillion-dollar fraud when a deepfake emulated a board member in a video allegedly approving a fraudulent transfer of funds. “Banks and financial institutions are particularly at risk,” said The Hack Academy. “A study by Deloitte found that over 50% of senior executives expect deepfake scams to target their organizations soon. These attacks can undermine trust and lead to significant financial loss.”  Hack Academy went on to say that AI-inspired security attacks weren’t confined to deepfakes. These attacks were also beginning to occur with increased regularity in the form of corporate espionage and misinformation campaigns. AI brings new, more dangerous tactics to traditional security attack methods like phishing, social engineering and the insertion of malware into systems. For CIOs, enterprise AI system developers, data scientists and IT network professionals, AI changes the rules and the tactics for security, given AI’s limitless potential for both good and bad. This is forcing a reset in how IT thinks about security against malicious actors and intruders. Related:How Bad Actors are Exploiting AI What exactly is IT up against? The AI tools that are available on the dark web and in public cyber marketplaces give security perpetrators a wide choice of AI weaponry. Also, IoT and edge networks now present much broader enterprise attack surfaces. Security threats can come in videos, phone calls, social media sites, corporate systems and networks, vendor clouds, IoT devices, network end points, and virtually any entry point into a corporate IT environment that electronic communications can penetrate. Here are some of the current AI-embellished security attacks that companies are seeing: Convincing deepfake videos of corporate executives and stakeholders that are intended to dupe companies in pursuing certain actions or transferring certain assets or funds. This deep faking also extends to voice simulations of key personnel that are left as voicemails in corporate phone systems.  Phishing and spearfishing attacks that send convincing emails (some with malicious attachments) to employees, who mistakenly open them because they think the sender is their boss, the CEO or someone else they perceive as trusted. AI supercharges these attacks because it can automate and send out a large volume of emails that hit many employee email accounts. That AI continues to “learn” with the help of machine learning so it can discover new trusted sender candidates for future attacks.   Related:Adaptive messaging that uses generative AI to craft messages to users that correct grammar and that “learn” from corporate communication styles so they can more closely emulate corporate communications that make them seem legitimate. Mutating code that uses AI to change malware signatures on the fly so antivirus detection mechanisms can be evaded. Data poisoning that occurs when a corporate or cloud provider’s AI data repository is injected by malware that alters (“poisons”) so the data produces erroneous and misleading results.  Fighting Back With Tech To combat these supercharged AI-based security threats, IT has number of tools, techniques and strategies it can consider. Fighting deepfakes. Deepfakes can come in the form of videos, voicemails and photos. Since deepfakes are unstructured data objects that can’t be parsed in their native forms like real data, there are new tools on the market that can convert these objects into graphical representations that can be analyzed to evaluate whether there is something in an object that should or shouldn’t be there. The goal is to confirm authenticity.  Related:Fighting phishing and spear phishing. A combination of policy and practice works best to combat phishing and spear phishing attacks. Both types of attacks are predicated on users being tricked into opening an email attachment that they believe is from a trusted sender, so the first line of defense is educating (and repeat-educating) users on how to handle their email. For instance, a user should notify IT if they receive an email that seems unusual or unexpected, and they should never open it. IT should also review its current security tools. Is it still using older security monitoring software that doesn’t include more modern technologies like observability, which can check for security intrusions or malware at more atomic levels?  Is IT still using IAM (identity access management) software to track user identities and activities at a top level in the cloud and on top and atomic levels on premises, or has it also added cloud identity entitlements management (CIEM), which gives it an atomic level view of  user accesses and activities in the cloud? Better yet, has IT moved to identity governance administration (IGA), which can serve as an over-arching umbrella for IAM and CIEM plugins, plus provide detailed audit reports and automated compliance across all platforms? Fighting embedded malware code. Malware can lie dormant in systems for months, giving a bad actor the option to activate it whenever the timing is right. It’s all the more reason for IT to augment its security staff with new skillsets, such as that of the “threat hunter,” whose job is to examine networks, data and systems on a daily basis, hunting down malware that might be lurking within, and destroying it before it activates. Fighting with zero-trust networks. Internet of Things (IoT) devices come into companies with little or no security because IoT suppliers don’t pay much attention to it and there is a general expectation that corporate IT will configure devices to the appropriate security settings. The problem is, IT often forgets to do this. There are also times when users purchase their own IoT gear, and IT doesn’t know about it. Zero-trust networks help manage this, because they detect and report on everything that is added, subtracted or modified on the network. This gives IT visibility into new, potential security breach points. A second step is to formalize IT procedures for IoT devices so that no IoT device is deployed without the device’s security first being set to corporate standards.  Fighting AI data poisoning. AI models, systems and data should be continuously monitored for accuracy. As soon as they show lowered levels of accuracy or produce unusual conclusions, the data repository, inflows and outflows should be examined for quality and non-bias of data. If contamination is found, the system should be taken down, the data sanitized, and the sources of the contamination traced, tracked and disabled. Fighting AI with AI. Most every security tool on the market today contains AI functionality to detect anomalies, abnormal data patterns and user activities. Additionally, forensics AI can dissect a security breach that does occur, isolating how it happened, where it originated from and what caused it. Since most sites don’t have on-staff forensics experts, IT will have to train staff in forensics skills. Fighting with regular audits and vulnerability testing. Minimally, IT vulnerability testing should be performed on a quarterly basis, and full security audits on an annual basis. If sites use cloud providers, they should request each provider’s latest security audit for review. An outside auditor can also help sites prepare for future AI-driven security threats, because auditors stay on top of the industry, visit many different companies, and see many different situations. An advanced knowledge of threats that loom in the future helps sites prepare for new battles. Summary AI technology is moving faster than legal rulings and regulations. This leaves most IT departments “on their own” to develop security defenses against bad actors who use AI against them.  The good news is that IT already has insights into how bad actors intend to use AI, and there are tools on the market that can help defensive efforts. What’s been missing is a proactive and aggressive battle plan from IT. That has to start now. 

When Rom Kosla, CIO at Hewlett Packard Enterprise (HPE), joined the technology giant in July 2023, the move represented a big shift in direction. Previously CIO at retailer Ahold Delhaize and CIO for enterprise solutions at PepsiCo, Kosla was a consumer specialist who wanted to apply his knowledge in a new sector. “I liked the idea of working in a different industry,” he says. “I went from consumer products to retail grocery. Moving into the tech industry was a bit nerve-wracking because the concept of who the customers are is different. But since I grew up in IT, I figured I’d have the ability to navigate my way through the company.” Kosla had previously worked as a project manager for Nestlé and spent time with the consultancy Deloitte. Now approaching two years with HPE, Kosla leads HPE’s technology strategy and is responsible for how the company harnesses artificial intelligence (AI) and data. He also oversees e-commerce, app development, enterprise resource planning (ERP) and security operations. “The role has exceeded my expectations,” he says. “When you’re a CIO at a multinational, like when I was a divisional CIO at PepsiCo, you’re in the back office. Whether it’s strategy, transformation or customer engagement, the systems are the enablers of that back-office effort. At HPE, it’s different because we are customer zero.” Kosla says he prefers the term “customer gold” because he wants HPE to develop high-quality products. In addition to setting the internal digital strategy, he has an outward-facing role providing expert advice to customers. That part of his role reminds him of his time at Deloitte. “Those are opportunities to flex my prior experience and capabilities, and learn how to take our products, enable them, and share best practices,” he says. “HPE is like any other company. We use cloud systems and software-as-a-service products, including Salesforce and others. But underneath, we have HPE powering a lot of the capabilities.” The press release announcing Kosla’s appointment in 2023 said HPE believed his prior experiences in the digital front-end and running complex supply chains made him the perfect person to build on its digital transformation efforts. So, how has that vision panned out? “What’s been interesting is helping the business and IT team think about the end-to-end value stream,” he says. “There was a lot of application-specific knowledge. The ability for processes to be optimised at an application layer versus the end-to-end value stream was only happening in certain spots.” Kosla discovered the organisation had spent two years moving to a private cloud installation on the company’s hardware and had consolidated 20-plus ERP systems under one SAP instance. With much of the transformation work complete, his focus turned to making the most of these assets. “The opportunity was not to shepherd up transformation, it was taking the next step, which was optimising,” says Kosla, explaining how he had boosted supply chain performance in his earlier roles. He’s now applying that knowledge at HPE. “What we’ve been doing is slicing areas of opportunity,” he says. “With the lead-to-quote process, for example, we have opportunities to optimise, depending on the type of business, such as the channel and distributors. We’re asking things like, ‘Can we get a quote out as quickly as possible, can we price it correctly, and can we rely less on human engagement?’” HPE announced a cost-reduction programme in March to reduce structural operating costs. The programme is expected to be implemented through fiscal year 2026 and deliver gross savings of approximately $350m by fiscal year 2027, including through workforce reductions. The programme of work in IT will help the company move towards these targets. Kosla says optimisation in financials might mean closing books faster. In the supply chain, the optimisation might be about predicting the raw materials needed to create products. He takes a term from his time in the consumer-packaged goods sector – right to play, right to win – to explain how his approach helps the business look for value-generating opportunities. “So, do we have the right to play, meaning do we have the skills? Where do we have the right to win, meaning do we have the funding, business resources and availability to deliver the results? We spend time focusing on which areas offer the right to play and the right to win.” Kosla says data and AI play a key role in these optimisations. HPE uses third-party applications with built-in AI capabilities and has developed an internal chat solution called ChatHPE, a generative AI hub used for internal processes. “There are lots of conversations around how we unlock the benefits of AI in the company,” he says. Professionals across the company use Microsoft Copilot in their day-to-day roles to boost productivity. Developers, meanwhile, use GitHub Copilot. Finally, there’s ChatHPE, which Kosla says is used according to the functional use case. HPE started developing the platform about 18 months ago. A pipeline of use cases has now been developed, including helping legal teams to review contracts, boosting customer service in operations, re-using campaign elements in marketing and improving analytics in finance. “We spend time focusing on which areas offer the right to play and the right to win” Rom Kosla, Hewlett Packard Enterprise “We have a significant amount of governance internally,” says Kosla, referring to ChatHPE, which is powered by Azure and OpenAI technology. “When I started, there wasn’t an internal HPE AI engine. We had to tell the teams not to use the standard tools because any data that you feed into them is ultimately extracted. So, we had to create our platform.” Embracing AI isn’t Kosla’s only concern. Stabilisation is a big part of what he needs to achieve during the next 12 months. He returns to HPE’s two major transformation initiatives – the shift to private cloud and the consolidation of ERP platforms – suggesting that the dual roll-out and management of these initiatives created a significant number of incidents. “When I look back at PepsiCo, we had about 300,000 employees and about 600,000 tickets, which means two tickets per person per year. I said to the executive committee at HPE, ‘We have 60,000 employees, and we have a couple of million tickets’, which is an insane number. The goal was to bring that number down by about 85%,” he says. “Now, our system uptime is 99% across our quoting and financial systems. That availability allows our business to do more than focus on internal IT. They can focus on the customer. Stabilisation means the business isn’t constantly thinking about IT systems, because it’s a challenge to execute every day when systems are going down because of issues.” Kosla says the long-term aim from an IT perspective is to align the technology organisation with business outcomes. In financials, for example, he wants to produce the data analytics the business needs across the supply chain and operational processes. “We have embedded teams that work together to look at how we enable data, like our chat capabilities, into some of the activities,” he says. “They’ll consider how we reduce friction, especially the manual steps. They’ll also consider planning, from raw materials to the manufacturing and delivery of products. That work involves partnering with the business.” The key to success for the IT team is to help the business unlock value quicker. “I would say that’s the biggest part for us,” says Kosla. “We don’t even like to use the word speed – we say velocity, because velocity equals direction, and that’s crucial for us. I think the business is happy with what we’ve been able to achieve, but it’s still not fast enough.” Being able to deliver results at pace will rely on new levels of flexibility. Rather than being wedded to a 12-month plan that maps out a series of deliverables, Kosla wants his team to work more in the moment. Prior experiences from the consumer sector give him a good sense of what excellence looks like in this area. “You don’t need to go back to the top, go through an annual planning review, go back down, and then have the teams twiddling their thumbs while they wait for the OK,” he says. “The goal is that teams are constantly working on what’s achievable during a sprint window. Many companies take that approach; I’ve done it in my prior working life. I know what can happen, and I think flexibility will drive value creation.” Kosla says some of the value will come from HPE’s in-house developed technologies. “One of the things that makes this role fun is that there’s a significant amount of innovation the company is doing,” he says, pointing to important technologies, such as Morpheus VM Essentials virtualisation software, the observability platform OpsRamp, and Aruba Networking Access Points. “What I’m proud of is that we now show up to customers with comparability,” he says, talking about the advisory part of his role. “We can say, ‘Look, we use both products, because in some cases, it’s a migration over time.’ So, for example, when a customer asks about our observability approach, we can compare our technology with other providers.” Kosla reflects on his career and ponders the future of the CIO role, suggesting responsibilities will vary considerably according to sector. “Digital leaders still maintain IT systems in some industries,” he says. “However, the rest of the business is now much more aware of technology. The blurring of lines between business and IT means it’s tougher to differentiate between the two areas. I think we’ll see more convergence.” Kosla says a growing desire to contain costs often creates a close relationship between IT and finance leaders. Once again, he expects further developments in that partnership. He also anticipates that cyber will remain at the forefront of digital leaders’ priority lists. More generally, he believes all IT professionals are becoming more focused on business priorities. “I think the blurring will continue to create interesting results, especially in technology companies,” he says. “We want to do things differently.” Read more interviews with tech company IT leaders Interview: Joe Depa, global chief innovation officer, EY – Accounting firm EY is focused on ‘AI-ready data’ to maximise the benefits of agentic AI and enable the use of emerging frontier technologies for its business and clients. Interview: Cynthia Stoddard, CIO, Adobe – After nearly 10 years in post, Adobe’s CIO is still driving digital transformation and looking to deliver lasting change through technology. Interview: Tomer Cohen, chief product officer, LinkedIn – The professional social network’s product chief is leading the introduction of artificial intelligence for the firm’s in-house development processes and to enhance services for users.

Google CEO Sundar Pichai talking about Google Beam, formerly known as Project Starline, at Google ... More I/O 2025Anshel Sag Google used its annual I/O event this week to put the focus squarely on AI — with a strong dash of XR. While there’s no doubt that Google remains very committed to Android and the Android ecosystem, it was more than apparent that the company’s work on AI is only accelerating. Onstage, Google executives showed how its Gemini AI models have seen a more than 50x increase in monthly token usage over the past year, with the major inflection point clearly being the release of Gemini 2.5 in March 2025. I believe that Google’s efforts in AI have been supercharged by Gemini 2.5 and the agentic era of AI. The company also showed its continued commitment to getting Android XR off the ground with the second developer preview of Android XR, which it also announced at Google I/O. (Note: Google is an advisory client of my firm, Moor Insights & Strategy.) Google’s monthly tokens processedAnshel Sag Incorporating Gemini And AI Everywhere For Google, the best way to justify the long-term and continuous investment in Gemini is to make it accessible in as many ways as possible. That includes expanding into markets beyond the smartphone and browser. That’s why Gemini is already replacing Google Assistant in most areas. This is also a necessary move because Google Assistant’s functionality has regressed to the point of frustration as the company has shifted development resources to Gemini. This means that we’re getting Gemini via Google TV, Android Auto and WearOS. Let’s not forget that Android XR is the first operating system from Google that has been built from the ground up during the Gemini era. That translates to most XR experiences from Google being grounded in AI from the outset to make the most of agents and multimodal AI for improving the user experience. To accelerate the pace of adoption of on-device AI, Google has also announced improvements to LiteRT, its runtime for using AI models locally that has a heavy focus on maximizing on-device NPUs. Google also announced the AI Edge Portal to enable developers to test and benchmark their on-device models. These models will be crucial for enabling low-latency and secure experiences for users when connectivity might be challenged or when data simply cannot leave the device. While I believe that on-device AI performance is going to be important to developers going forward, it is also important to recognize that hybrid AI — mixing on-device and cloud AI processing — is likely here to stay for a very long time. Android XR, Smart Glasses And The Xreal Partnership Because Google introduced most of its Android updates in a separate “Android Show” a week before Google I/O, the Android updates during I/O mostly applied to Android XR. The new Material 3 Expressive design system will find its way across Google’s OSes and looks set to deliver snappier, more responsive experiences at equal or better performance. I wrote extensively about Google’s Android XR launch in December 2024, explaining how it would likely serve as Google’s tip of the spear for enabling new and unique AI experiences. At Google I/O, the company showed the sum of these efforts in terms of both creating partnerships and enabling a spectrum of XR devices from partners.Google’s Shahram Izadi, vice president and general manager of Android XR, talking about Project ... More Moohan onstage at Google I/O 2025Anshel Sag In this vein, Google reiterated its commitment to Samsung and Project Moohan, which Google now says will ship this year. The company also talked about other partnerships in the ecosystem that will enable new form factors for the AI-enabled wearable XR operating system. Specifically, it will be partnering with Warby Parker and Gentle Monster to develop smart glasses. In a press release, Google said it has allotted $150 million for its partnership with Warby Parker, with $75 million already committed to product development and commercialization and the remaining $75 million dependent on reaching certain milestones. I believe that this partnership is akin to the one that Meta established with EssilorLuxottica, leaving the design, fit and retail presence to the eyeglasses experts. Warby Parker is such a good fit because the company is already very forward-thinking on technology, and I believe that this partnership can enable Google to make some beautiful smart glasses to compete with Meta Ray Bans. While I absolutely adore my Meta Ray Bans, I do think they would be considerably more useful if they were running Gemini 2.5, even the flash version of the model. Gentle Monster is also a great fit for Google because it helps capture the Asian market better, and because its designs are so large that they give Google plenty of room to work with. Many people have written about their impressions of Project Moohan and the smart glasses from Google I/O, but the reality is that these were not new — or final — products. So, I hope that these XR devices are as exciting to people as they were to me back in December.Google announces Project Aura on stage during the Google I/O developer keynote.Anshel Sag For me the more important XR news from the event was the announcement of the Project Aura headset in partnership with Xreal. Project Aura, while still limited in details, does seem to indicate that there’s a middle ground for Google between the more immersive Moohan headset and lightweight smart glasses. It’s evident that Google wants to capture this sweet spot with Xreal’s help. Also, if you know anything about Xreal’s history, it makes sense that it would be the company Google works with to bring 3-D AR to market. Project Aura feels like Google’s way to compete with Meta’s Orion in terms of field of view, 3-D AR capabilities and standalone compute. While many people think of Orion as a pair of standalone glasses, in fact they depend on an external compute puck; with Qualcomm’s help, Google will also use a puck via a wire, though I would love to see that disappear in subsequent versions. The Xreal One and One Pro products already feel like moves in the direction Google is leaning, but with Project Aura it seems that Google wants more diversity within Android XR — and it wants to build a product with the company that has already shipped more AR headsets than anyone else. The wider 70-degree field of view should do wonders for the user experience, and while the price of Project Aura is still unclear, I would expect it to be much more expensive than most of Xreal’s current offerings. Google and Xreal say they will disclose more details about Project Aura at the AWE 2025 show in June, which I will be attending — so look for more details from me when that happens. Project Starline Becomes Google Beam Google also updated its XR conferencing platform, formerly called Project Starline, which it has been building with HP. Google has now changed the project into a product name with the introduction of Google Beam. While not that much has changed since I last tried out Project Starline at HP’s headquarters last September, the technology is still quite impressive — and still quite expensive. One of the new capabilities for Google Beam, also being made available as part of Google Meet, is near-real-time translated conversations that capture a person’s tone, expressions and accents while translating their speech. I got to experience this at Google I/O, and it was extremely convincing, not to mention a great way to enhance the already quite impressive Beam experience. It really did sound like the translated voice was the person’s own voice speaking English; this was significant on its own, but achieving it with spatial video at fairly low latency was even better. I hope that Google will one day be able to do the translations in real time, synced with the user’s speech. Google says that it and HP are still coming to market with a Google Beam product later this year and will be showing it off at the InfoComm conference in June. Google has already listed some lead customers for Google Beam, including Deloitte, Salesforce, Citadel, NEC, Hackensack Meridian Health, Duolingo and Recruit. This is a longer list than I expected, but the technology is also more impressive than I had initially expected, so I am happy to see it finally come to market. I do believe that with time we’ll probably see Google Beam expand beyond the 65-inch screen, but for now that’s the best way to attain full immersion. I also expect that sooner or later we could see Beam working with Android XR devices as well. Analyst Takeaways From Google I/O I believe that Google is one of the few companies that genuinely understands the intersection of AI and XR — and that has the assets and capabilities to leverage that understanding. Other companies may have the knowledge but lack the assets, capabilities or execution. I also believe that Google finally understands the “why” behind XR and how much AI helps answer that question. Google’s previous efforts in XR were for the sake of pursuing XR and didn’t really align well with the rest of the company’s efforts. Especially given the growth of AI overall and the capabilities of Gemini in particular, AR glasses are now one of the best ways to experience AI. Nobody wants to hold their phone up to something for a multimodel AI to see it, and no one wants to type long AI prompts into their phone. They want to interact with AI in the context of more natural visual and auditory experiences. Although smartphones can deliver a fairly good experience for this, they pale in comparison to having the microphones and cameras closer to your eyes and mouth. The more you use AI this way, the less you find yourself needing to pull out your phone. I certainly don’t think smartphones are going to disappear, but I do think they are going to decline in terms of where most of an individual’s AI computing and connectivity happen. All of this is why I’m much more confident in Google’s approach to XR this time around, even though the company has burned so many bridges with its previous endeavors in the space (specifically Daydream and Glass). More than that, I believe that Google’s previous absence in the XR market has impeded the market’s growth. Now, however, the company is clearly investing in partnerships and ecosystem enablement. It will be important for the company to continue to execute on this and enable its partners to be successful. A big part of that is building a strong XR ecosystem that can compete with the likes of Apple and Meta. It won’t happen overnight, but the success of that ecosystem will be what makes or breaks Google’s approach to XR beyond its embrace of Gemini. Moor Insights & Strategy provides or has provided paid services to technology companies, like all tech industry research and analyst firms. These services include research, analysis, advising, consulting, benchmarking, acquisition matchmaking and video and speaking sponsorships. Of the companies mentioned in this article, Moor Insights & Strategy currently has (or has had) a paid business relationship with Google, HP, Meta, Qualcomm, Salesforce and Samsung.Editorial StandardsReprints & Permissions

Deloitte UK is reducing some bonuses. Peter Dazeley via Getty Images 2025-05-21T16:42:47Z Save Saved Read in app This story is available exclusively to Business Insider subscribers. Become an Insider and start reading now. Have an account? Staff in one of Deloitte's four UK divisions will get lower bonuses after missing profit targets. The firm will also promote fewer staff this year, according to an internal email seen by BI. The promotions move was demotivating and has "dampened" the office mood, a Deloitte employee said. Deloitte told employees in one of its UK consulting divisions to expect lower annual bonuses following weaker-than-expected financial performance.Richard Houston, a senior partner and chief executive of Deloitte UK, wrote in a companywide email on Tuesday that employees in the technology and transformation (T&T) consulting division would receive an average of 80% of their annual bonus.Partners in the division will also face a drop in annual rewards, Houston said in the memo seen by Business Insider.The news was first reported by The Financial Times.The T&T business had "faced a particularly challenging year and fell materially short of its performance goals," Houston wrote. Richard Houston is chief executive of Deloitte UK. Liam McBurney/PA/Getty Images Bonuses have always been linked to individual and firm performance. However, for the 2025 financial year, they will also take into account performance variations in each of Deloitte's four business lines.The shift means that workers in Deloitte's other three business lines — which cover deals, tax and legal, and audit and assurance services — will retain their full bonus.Those three divisions have either exceeded targets or, in the case of the audit and assurance branch, had still increased profits in the 2025 financial year, Houston wrote.Deloitte will also promote fewer employees than in previous years, Houston said.The firm would promote 5,500 employees, or about 25% of the UK workforce, at the start of the next financial year, he said. The previous year 28% were promoted, Houston said.A Deloitte consultant in the T&T division said the news about promotions was demotivating and had "dampened" the mood. The person added that their division had taken "a beating."It's "not the best feeling since T&T functions have historically pulled the firm forward in times of crisis," the employee added.Deloitte UK's total profits for its 2025 financial year, which ends on May 31, would be "slightly ahead of last year" but "below our original plan," Houston wrote."At the start of FY25, we expected greater economic stability and a gradual return of growth opportunities. But an early election, geopolitical complexity, and unexpected economic headwinds — like changes in trade policies — have continued to cause market uncertainty," he said.Deloitte's global revenue climbed 3.1% to $67.2 billion in the 2024 financial year, which marked a sharp slowdown in growth compared to the 14.9% increase in 2023.The firm's UK arm has reorganized its business divisions and laid off workers as it grappled with an industrywide slowdown in demand for consulting services that has hit revenue growth.Cost-cuttingIn October 2024, Deloitte UK cut its spending on staff travel and expenses by more than 50%.Houston addressed the cost-cutting measures in Tuesday's email, acknowledging "it has not been easy" for teams to be unable to meet in person for much of the year.He said the efforts to manage costs had "made a significant difference" and had directly contributed to bonuses for the 2025 financial year.A Deloitte UK spokesperson told BI: "Amid ongoing market uncertainty, we are pleased to be able to recognize our people for their hard work with salary increases, bonuses, and promotions this year. This is alongside other benefits such as fully funded private medical insurance, recently enhanced family policies, and our commitment to offering flexibility and choice in our ways of working."Have a tip? Contact this reporter via email at pthompson@businessinsider.com or Signal at Polly_Thompson.89. Use a personal email address and a nonwork device; here's our guide to sharing information securely.

Google just wrapped up its big keynote at I/O 2025. As expected, it was full of AI-related announcements, ranging from updates across Google’s image and video generation models to new features in Search and Gmail.But there were some surprises, too, like a new AI filmmaking app and an update to Project Starline. If you didn’t catch the event live, you can check out everything you missed in the roundup below.Google’s AI Mode for Search is coming to everyoneGoogle has announced that it’s rolling out AI Mode, a new tab that lets you search the web using the company’s Gemini AI chatbot, to all users in the US starting this week.Google will test new features in AI Mode this summer, such as deep search and a way to generate charts for finance and sports queries. It’s also rolling out the ability to shop in AI Mode in the “coming months.”Project Starline is now Google BeamImage: GoogleProject Starline, which began as a 3D video chat booth, is taking a big step forward. It’s becoming Google Beam and will soon launch inside an HP-branded device with a light field display and six cameras to create a 3D image of the person you’re chatting with on a video call.Companies like Deloitte, Duolingo, and Salesforce have already said that they will add HP’s Google Beam devices to their offices.Imagen and Veo are getting some big upgradesGoogle has announced Imagen 4, the latest version of its AI text-to-image generator, which the company says is better at generating text and offers the ability to export images in more formats, like square and landscape. Its next-gen AI video generator, Veo 3, will let you generate video and sound together, while Veo 2 now comes with tools like camera controls and object removal.Google launches an AI filmmaking appGIF: GoogleIn addition to updating its AI models, Google is launching a new AI filmmaking app called Flow. The tool uses Veo, Imagen, and Gemini to create eight-second AI-generated video clips based on text prompts and / or images. It also comes with scene-builder tools to stitch clips together and create longer AI videos.Gemini 2.5 Pro adds an “enhanced” reasoning modeThe experimental Deep Think mode is meant for complex queries related to math and coding. It’s capable of considering “multiple hypotheses before responding” and will only be available to trusted testers first.Google has also made its Gemini 2.5 Flash model available to everyone on its Gemini app and is bringing improvements to the cost-efficient model in Google AI Studio ahead of a wider rollout.Xreal shows off its Project Aura prototypeXreal and Google are teaming up on Project Aura, a new pair of smart glasses that use the Android XR platform for mixed-reality devices. We don’t know much about the glasses just yet, but they’ll come with Gemini integration and a large field-of-view, along with what appears to be built-in cameras and microphones.Google is also partnering with Samsung, Gentle Monster, and Warby Parker to create other Android XR smart glasses, as well.Google’s experimental AI assistant is getting more proactive Project Astra could already use your phone’s camera to “see” the objects around you, but the latest prototype will let it complete tasks on your behalf, even if you don’t explicitly ask it to. The model can choose to speak based on what it’s seeing, such as pointing out a mistake on your homework.Gemini is coming to ChromeGoogle is building its AI assistant into Chrome. Starting on May 21st, Google AI Pro and Ultra subscribers will be able to select the Gemini button in Chrome to clarify or summarize information across webpages and navigate sites on their behalf. The feature can work with up to two tabs for now, but Google plans on adding support for more later this year.Google’s new AI Ultra plan costs $250 per monthGoogle is rolling out a new “AI Ultra” subscription that offers access to the company’s most advanced AI models and higher usage limits across apps like Gemini, NotebookLM, Flow, and more. The subscription also includes early access to Gemini in Chrome and Project Mariner, which can now complete up to 10 tasks at once.Search Live will let you discuss what’s on your camera in real-timeSpeaking of Project Astra, Google is launching Search Live, a feature that incorporates capabilities from the AI assistant. By selecting the new “Live” icon in AI Mode or Lens, you can talk back and forth with Search while showing what’s on your camera.After making Gemini Live’s screensharing feature free for all Android users last month, Google has announced that iOS users will be able to access it for free, as well.Image: GoogleGoogle has revealed Stitch, a new AI-powered tool that can generate interfaces using selected themes and a description. You can also incorporate wireframes, rough sketches, and screenshots of other UI designs to guide Stitch’s output. The experiment is currently available on Google Labs.Google Meet adds AI speech translationImage: GoogleGoogle Meet is launching a new feature that translates your speech into your conversation partner’s preferred language in near real-time. The feature only supports English and Spanish for now. It’s rolling out in beta to Google AI Pro and Ultra subscribers.Gmail’s smart replies will soon pull info from your inboxImage: GoogleGmail’s smart reply feature, which uses AI to suggest replies to your emails, will now use information from your inbox and Google Drive to prewrite responses that sound more like you. The feature will also take your recipient’s tone into account, allowing it to suggest more formal responses in a conversation with your boss, for example.Gmail’s upgraded smart replies will be available in English on the web, iOS, and Android when it launches through Google Labs in July.Image: GoogleGoogle is testing a new feature that lets you upload a full-length photo of yourself to see how shirts, pants, dresses, or skirts might look on you. It uses an AI model that “understands the human body and nuances of clothing.”Google will also soon let you shop in AI Mode, as well as use an “agentic checkout” feature that can purchase products on your behalf.Google Chrome will soon help you update compromised passwordsIf Chrome detects that your password’s been compromised, Google says the browser will soon be able to “generate a strong replacement” and automatically update it on supported websites. The feature launches later this year, and Google says that it will always ask for consent before changing your passwords.See More: