In the wake of parting ways with gaming personality Dr. Disrespect, Midnight Society has closed its doors and killed the first-person shooter game Dead Drop.Read More

The rapid advancement of Large Language Models (LLMs) has significantly improved their ability to generate long-form responses. However, evaluating these responses efficiently and fairly remains a critical challenge. Traditionally, human evaluation has been the gold standard, but it is costly, time-consuming, and prone to bias. To mitigate these limitations, the LLM-as-a-Judge paradigm has emerged, leveraging LLMs themselves to act as evaluators. Despite this advancement, LLM-as-a-Judge models face two significant challenges: (1) a lack of human-annotated Chain-of-Thought (CoT) rationales, which are essential for structured and transparent evaluation, and (2) existing approaches that rely on rigid, hand-designed evaluation components, making them difficult to generalize across different tasks and domains. These constraints limit the accuracy and robustness of AI-based evaluation models. To overcome these issues, Meta AI has introduced EvalPlanner, a novel approach designed to improve the reasoning and decision-making capabilities of LLM-based judges through an optimized planning-execution strategy.EvalPlanner is a preference optimization algorithm specifically designed for Thinking-LLM-as-a-Judge models. EvalPlanner differentiates itself by employing a three-stage evaluation process: (1) generation of an unconstrained evaluation plan, (2) execution of the plan, and (3) final judgment. Unlike previous methods, EvalPlanner does not constrain reasoning traces to predefined rubrics or criteria. Instead, it generates flexible evaluation plans that adapt to various domains and task requirements. The system operates in a self-training loop, iteratively refining evaluation plans and execution strategies using synthetically generated preference pairs. By continuously optimizing itself, EvalPlanner ensures more reliable, transparent, and scalable evaluations compared to existing LLM-as-a-Judge models.The innovation behind EvalPlanner lies in its structured reasoning approach, which separates the planning phase from the execution phase. In the planning stage, the model formulates a detailed evaluation roadmap tailored to the specific instruction at hand. During execution, the model follows the step-by-step plan to assess and compare responses systematically. This two-step separation enables better alignment between evaluation goals and reasoning processes, leading to more accurate and explainable judgments.Technical Details and Benefits of EvalPlannerEvalPlanner introduces a self-training mechanism that continuously refines both the planning and execution components of the evaluation process. The model leverages Direct Preference Optimization (DPO) to iteratively improve its judgments by learning from synthetic preference pairs. These preference pairs are derived by sampling multiple evaluation plans and executions, allowing EvalPlanner to identify the most effective reasoning patterns.The primary benefits of EvalPlanner include:Increased Accuracy: By generating unconstrained evaluation plans, EvalPlanner significantly reduces bias and improves judgment consistency across different tasks.Scalability: Unlike manually crafted evaluation rubrics, EvalPlanner automatically adapts to new evaluation tasks, making it a highly scalable solution.Efficiency: EvalPlanner achieves state-of-the-art (SOTA) performance on various benchmarks with fewer training examples, relying only on synthetic preference pairs rather than extensive human annotations.Transparency: By explicitly separating planning from execution, EvalPlanner enhances the interpretability of its reasoning process, making it easier to analyze and debug.Experimental Results and Performance InsightsMeta AI evaluated EvalPlanner across multiple reward modeling benchmarks, including RewardBench, RM-Bench, JudgeBench, and FollowBenchEval. The results demonstrate EvalPlanners superior performance in evaluating complex, multi-level constraints and improving over existing models in various domains, such as chat-based interactions, safety evaluation, coding, and mathematical reasoning.State-of-the-Art Results on RewardBench: EvalPlanner achieved a score of 93.9, outperforming leading models that rely on 30 times more human-annotated data. This highlights the effectiveness of EvalPlanners synthetic data-driven training methodology.Improved Robustness on RM-Bench: EvalPlanner demonstrated 8% higher accuracy compared to previous SOTA models in handling nuanced evaluation criteria, showcasing its ability to resist subtle biases and variations in response quality.Superior Constraint Handling in FollowBenchEval: For multi-level constraints evaluation, EvalPlanner outperformed competitive baselines by 13%, emphasizing its ability to effectively plan and reason through complex prompts.Generalization to JudgeBench: EvalPlanner demonstrated strong generalization capabilities, achieving comparable performance to larger models trained on extensive human-annotated datasets while using significantly fewer preference pairs.Additionally, ablation studies confirmed that iterative optimization of evaluation plans significantly enhances performance. When trained with as few as 5K synthetic preference pairs, EvalPlanner maintained competitive performance, demonstrating its data efficiency compared to traditional models.Conclusion: The Future of AI-Based EvaluationEvalPlanner represents a major breakthrough in the development of AI-based evaluation frameworks. By combining preference optimization, structured planning, and self-training, it effectively addresses the limitations of existing LLM-as-a-Judge models. Its scalability, accuracy, and transparency make it a promising tool for automated, unbiased, and efficient evaluation of AI-generated responses across diverse applications. As AI models continue to evolve, EvalPlanner paves the way for more reliable and interpretable evaluation systems, ultimately enhancing trust and fairness in AI-driven decision-making. Future research can explore extending EvalPlanners capabilities to reward modeling in Reinforcement Learning with Human Feedback (RLHF) pipelines and integrating it into real-world AI auditing frameworks.With EvalPlanner, Meta AI has set a new standard in the field of AI evaluation, demonstrating that teaching AI to plan and reason can significantly improve judgment quality. This advancement is a crucial step toward autonomous and scalable AI governance, ensuring that future AI systems operate with greater precision, fairness, and accountability.Check out the Paper. All credit for this research goes to the researchers of this project. Also,dont forget to follow us onTwitter and join ourTelegram Channel andLinkedIn Group. Dont Forget to join our70k+ ML SubReddit.(Promoted) Asif RazzaqWebsite| + postsBioAsif Razzaq is the CEO of Marktechpost Media Inc.. As a visionary entrepreneur and engineer, Asif is committed to harnessing the potential of Artificial Intelligence for social good. His most recent endeavor is the launch of an Artificial Intelligence Media Platform, Marktechpost, which stands out for its in-depth coverage of machine learning and deep learning news that is both technically sound and easily understandable by a wide audience. The platform boasts of over 2 million monthly views, illustrating its popularity among audiences.Asif Razzaqhttps://www.marktechpost.com/author/6flvq/Yandex Develops and Open-Sources Perforator: An Open-Source Tool that can Save Businesses Billions of Dollars a Year on Server InfrastructureAsif Razzaqhttps://www.marktechpost.com/author/6flvq/NVIDIA AI Releases Eagle2 Series Vision-Language Model: Achieving SOTA Results Across Various Multimodal BenchmarksAsif Razzaqhttps://www.marktechpost.com/author/6flvq/Qwen AI Releases Qwen2.5-VL: A Powerful Vision-Language Model for Seamless Computer InteractionAsif Razzaqhttps://www.marktechpost.com/author/6flvq/DeepSeek-AI Releases Janus-Pro 7B: An Open-Source multimodal AI that Beats DALL-E 3 and Stable Diffusion [Recommended] Join Our Telegram Channel

This article contains spoilers for Severance season 2 episode 3. The long wait between seasons for sci-fi thriller Severance created a sense of urgency among viewers to get back up to date. Apple TV+ led off Severance season 2 with a recap of the previous seasons events. Meanwhile, the digital media industry embarked on a Severance Season 1 Recap goldrush, with some nerd at Den of Geek even penning a previously on primer.Still, despite that three year delay, Severance season 2 has proven to be relatively easy to follow thus far. Episodes 1 and 2 each picked up right after the unforgettable season 1 finale, and charted the immediate aftermath of the Overtime Contingency and its implications for the innie and outtie worlds. Its not until this weeks third episode, however, that the recollective momentum falls off the tracks a bit. Thats because Severance season 2 episode 3 Who Is Alive? reintroduces an important character from season 1 named Reghabi who you almost certainly forgot all about. Who is Asal Reghabi and what is this reintegration she speaks of? Allow us to reintegrate you on both fronts.Who Is Asal Reghabi?Played by Karen Aldridge, Asal Reghabi is a key figure on Severance (and not just because her name sounds the most like a Star Wars character of anyone in the cast). First introduced in season 1 episode 6 Hide and Seek, Reghabi was a Lumon employee who went rogue. She appears to have worked for the company in a scientific or biomedical capacity as she claims to have installed the severance chip in several employees brains including in Marks (Adam Scott).Now she works against Lumon by developing experimental reintegration technology that will un-do the severance process and make severed employees whole once again. Its because of this reintegration work that her path crosses with Mark. Early on in season 1, outtie Mark is contacted by his old work friend Petey Kilmer (Yul Vazquez). Petey claims to be undergoing the reintegration procedure and wants Marks help in joining a resistance against Lumon. Unfortunately Petey eventually dies from complications of reintegration.When Mark turns on a cellphone that Petey left behind, he immediately receives a phone call from Reghabi and agrees to meet with her in an old lab building at Ganz College. Before Mark can get much information, he is approached by Lumon thug Doug Graner (Michael Cumpsty) who Reghabi promptly beats to death with a baseball bat. She instructs Mark to go home while she handles Graners body and assures him that theyll meet back up again at some point to finish what Petey started.We were interrupted, if you recall, Reghabi tells Mark in this episode of their first meeting. Indeed we do now recall.What Is Reintegration?Reghabi arrives to Marks home in Severance season 2 and convinces him not do something dumb by burning a message on his retinas for his innie to read. Apparently the switch from outtie to innie dilates the pupils to provide a clean visual slate anyway. Instead, Reghabi advocates for Mark to do something arguably even dumber to get information in and out of Lumon: attempt reintegration.As Reghabi tells it, reintegration is a way to sew together the innie and outtie consciousnesses. We saw a little how that worked in season 1 when Petey, despite quitting the MDR floor on Lumon, was still able to access some fractured memories from the inside. Of course, Petey also paid dearly for those memories, something that Reghabi is eager to explain away to Mark.When she first met Mark, Reghabi told him that Petey would have survived reintegration if he had followed her instructions and not panicked the moment he experienced adverse side effects. Here, she tries to assure him that shes fine-tuned the process even further. But what exactly is that process? It looks quite similar to the sci-fi ritual of the severance procedure itself. Join our mailing listGet the best of Den of Geek delivered right to your inbox!The monitors differentiate the five brainwave frequencies of the innie and the outtie: delta, theta, alpha, beta, gamma, Reghabi says to Mark of the machine hes hooked up to. One frequency, two waves per oscilloscope. The waves arent in sync. Not yet, anyway.The interesting part, however, comes when Reghabi fiddles with some knobs and asks Mark questions like Who am I? (Asal Reghabi), What was your mothers name? (Fern Scout), and What is something for which you feel shame? (Left the gate open and the family dog died). Its not until the What month is it? question that it becomes clear that reintegration is beginning to take hold.You mean what quarter? some semblance of innie Mark responds through outtie Marks mouth. One brilliantly-edited scene that finds Mark splicing into the Lumon conference room in his pajamas later and Marks reintegration journey has officially begun. One can only hope that Reghabi has indeed worked out all the kinks.The first three episodes of Severance season 2 are available to stream on Apple TV+ now. New episodes premiere Fridays, culminating with the finale on March 21.

Broadcom has released security updates to patch five security flaws impacting VMware Aria Operations and Aria Operations for Logs, warning customers that attackers could exploit them to gain elevated access or obtain sensitive information.The list of identified flaws, which impact versions 8.x of the software, is below -CVE-2025-22218 (CVSS score: 8.5) - A malicious actor with View Only Admin permissions may be able to read the credentials of a VMware product integrated with VMware Aria Operations for LogsCVE-2025-22219 (CVSS score: 6.8) - A malicious actor with non-administrative privileges may be able to inject a malicious script that may lead to arbitrary operations as admin user via a stored cross-site scripting (XSS) attackCVE-2025-22220 (CVSS score: 4.3) - A malicious actor with non-administrative privileges and network access to Aria Operations for Logs API may be able to perform certain operations in the context of an admin userCVE-2025-22221 (CVSS score: 5.2) - A malicious actor with admin privileges to VMware Aria Operations for Logs may be able to inject a malicious script that could be executed in a victim's browser when performing a delete action in the Agent ConfigurationCVE-2025-22222 (CVSS score: 7.7) - A malicious user with non-administrative privileges may exploit this vulnerability to retrieve credentials for an outbound plugin if a valid service credential ID is knownSecurity researchers Maxime Escourbiac from Michelin CERT, and Yassine Bengana and Quentin Ebel from Abicom and part of the Michelin CERT team for detecting and reporting the flaws. It's worth noting that the same team spotted two other shortcomings in the same product (CVE-2024-38832 and CVE-2024-38833) in late November 2024.All the aforementioned vulnerabilities have been patched in VMware Aria Operations and Aria Operations for Logs version 8.18.3. The virtualization services provider makes no mention of these issues being exploited in the wild.The advisory comes days after Broadcom warned of a high-severity security flaw in VMware Avi Load Balancer (CVE-2025-22217, CVSS score: 8.6) that could be weaponized by malicious actors to gain database access.Found this article interesting? Follow us on Twitter and LinkedIn to read more exclusive content we post.

Our client, a $25M+ AI product, AD agency, and professional services company, is seeking a visionary and experienced General Manager of Products and Services to lead its high-performing team and drive operational excellence. Reporting directly to the CEO, the General Manager will oversee all product and service departments, driving a culture of collaboration, accountability, and innovation.This role is a critical step toward the organization's future leadership, as the selected candidate will be included on the CEO succession slate and is expected to develop into the CEO position within 3-5 years.RequirementsStrategic Leadership: Provide visionary leadership, aligning teams and operations with the company's strategic goals.Operational Excellence: Develop and implement operational plans to optimize efficiency and deliver exceptional results across departments.Team Development: Mentor and manage direct reports, building a cohesive and high-performing team while fostering professional growth.Financial Oversight: Manage budgets, monitor performance metrics, and ensure all operational deliverables are met or exceeded.Long-Term Strategy: Collaborate with senior leadership to identify growth opportunities and develop innovative strategies for future success.Qualifications:Proven track record in senior or general management roles, preferably in the products and services industry.Exceptional leadership skills with demonstrated success in driving organizational results.Strong business acumen and the ability to manage complex operations effectively.Outstanding communication and interpersonal skills, with the ability to inspire and influence others.A bachelor's degree in business, management, or a related field is required; an advanced degree is preferred.Experience managing confidential processes with discretion is essential.BenefitsHealth Care Plan (Medical, Dental & Vision)Retirement Plan (401k, IRA)Life Insurance (Basic, Voluntary & AD&D)Paid Time Off (Vacation, Sick & Public Holidays)Family Leave (Maternity, Paternity)Short Term & Long Term DisabilityTraining & DevelopmentWork From Home

De Beauvoir Estate IX (2020) Source:&nbsp David GrandorgeWe must acknowledge that the great post-war social housing estates, despite their flaws, are still held in great affection by many who live in them, says David Grandorge This photograph was taken from a parking area (for residents use only) on the De Beauvoir Estate in Hackney, east London, in February 2020.It depicts, at its centre, an 18-storey tower in the mid-distance. On the right are the lower storeys of a tower of similar design and on the left a four-storey rendered building containing stacked maisonettes. A single-storey storage building or plant room (it was labelled vaguely) is attached to it. A narrow, inclined road beside this small structure gives vehicles access to the road above.The De Beauvoir Estate was completed in 1968, a year in which 420,000 new homes were built in the UK, many of them high-rise and system-built. In that same year, the recently completed system-built tower block Ronan Point suffered a partial collapse, a tragic event caused by a gas explosion.AdvertisementThis failure of machine, and consequently building fabric, led in the following years to widespread popular dissatisfaction with, and mistrust of, post-war housing. Two generations later, some of this mistrust lingers. But there are many who have lived in these housing types over an extended period who have not only got used to the austere language of the buildings they inhabit, but have come to embrace the collective way of life the architecture supports.The photograph above was one of many to be featured in an exhibition given the title Great Estates: An Incomplete Anthology of Social Housing in London. It was to be held in Stephen Taylors Building Workshop in the heart of the De Beauvoir Estate and was due to open on 19 March 2020. It was indefinitely postponed due to the spread of a virus that had a significant effect on, well, so many things.The pictures that were to be featured in this show echoed, in composition and subject matter, many of those taken by the German photographer Axel Htte of social housing in London. Made between 1982 and 1984, they were published in a book given the intentionally simple title London.Htte photographed many collective housing types from different eras in the districts of Spitalfields, Shoreditch, Hoxton, Bethnal Green, Shadwell and Limehouse in the East End and Bermondsey, Borough, Lambeth, Kennington and Camberwell in south London. The subjects were chosen with great care. He seemed to understand the DNA of the city at that time. Some of the examples he documented exist in a similar state to that shown in his very precise photographs. Others have been changed significantly by what has been built beside or beyond them. All have survived, with only minor changes to their external appearance mainly the addition of elements to make them more secure.AdvertisementThe world imagined by the estates designers did not materialise, but we must acknowledge that these great estates, despite their flaws, are still held in great affection by many, if not all, of their occupants. Local authorities take note. No more demolition please.David Grandorge is a photographer and senior lecturer in architecture at London Met. His fee for this column has been donated to support the publication of new and diverse voices in the AJ2025-01-31David Grandorgecomment and share

Initially approached by the client to add a side-return extension to the property, THISS Studio proposed instead to work with the existing building to provide a lower-cost, lower-carbon solution.The Victorian house, which previously featured enclosed spaces and dark interiors, has been opened up with a reconfigured floor plan that makes full use of the space that was already available.To provide cooking facilities for a large family and to enhance the connection to the garden, the practice created a spacious kitchen at the back of the house.AdvertisementA large void was discovered under the floor of this part of the house at the start of the project, which was exploited to add 1m of ceiling height. The new kitchen is now set a level down from the rest of the home and accessed by tile-lined steps.In one corner, a cantilevered dining bench borrows additional space from the garden and is framed by three large windows to help brighten the room.The kitchen and dining area, designed to be heart of the home, features pine timber furnishings, paired with terracotta floor tiles and pale cream acoustic wall panels made from recycled paper waste, making for a highly textured, tactile space.Working within the homes existing footprint and avoiding the need for a carbon-intensive, costly extension freed up budget for finishes and furnishings, such as the bespoke floor-to-ceiling mint green shelving unit and flower-shaped light fittings.Outside the kitchen, an aluminium canopy projects outwards from the corner of the building to add a sculptural feel to the exterior. A smaller curved aluminium ledge beneath the window mirrors the shape of the canopy above and doubles as a seat or table.AdvertisementRenovated spaces elsewhere on the ground floor include the front of the home, which has been converted from a kitchen into a yellow-painted living and study space, while a small combined bathroom and utility room now sits at the centre of the plan off the hallway.Architects viewBuilding bigger does not always mean youll have a space with functionality and quality. We worked with our clients to understand what they really needed as a family, which was actually better, more usable space. In rethinking the home as a team, we have saved a huge amount of carbon and allowed our clients budget to be redirected into quality, more sustainable materials and fittings that means their home has a sense of beauty, and they will love being there for many years to come.A creative reconfiguration of the existing floor plan has created a much-loved, carefully tailored home without the need for an extension, showing that sometimes unlocking the space already in our homes can be just as valuable as extending, with a fraction of the carbon.Sash Scott, founder, THISS StudioClients viewTHISS Studio has done a terrific job in opening up the space to suit our familys needs. The two front rooms have a really beautiful feel and so much more practical space, serving now as an office and family room. The kitchen, previously very dark, is now light, airy and uplifting. We really wanted it to be a sociable and convivial hub, a place where we could cook as well as socialise. A built-in bench seat allows for a generously-sized table, surrounded by views of the garden and trees through the beautiful windows. The natural timber and wall and ceiling panels add warmth and character.Sash and the team fully grasped the core of our vision, helping us to realise we didnt need to build outwards to create more functional space. The outcome is so special and different; the careful rethinking of space has avoided the environmental impact of an extension, instead creating something better and more beautiful within a footprint we could afford.Project dataLocation Waltham Forest, LondonStart on site May 2023Completion May 2024Gross internal floor area 64m2Construction cost UndisclosedArchitect THISS StudioClient PrivateInterior design THISS StudioStructural Engineer Detail SD

Looking forthe most recentMini Crossword answer?Click here for today's Mini Crossword hints, as well as our daily answers and hints for The New York Times Wordle, Strands, Connections and Connections: Sports Edition puzzles.The NYT Mini Crosswordwasn't too tough today. Movie fans, you should do pretty well, thanks to 7-Across and 8-Across. Need some more help with today's Mini Crossword? Read on. And if you could use some hints and guidance for daily solving, check out our Mini Crossword tips.The Mini Crossword is just one of many games in the Times' games collection. If you're looking for today's Wordle, Connections, Connections: Sports Edition and Strands answers, you can visitCNET's NYT puzzle hints page.Read more: Tips and Tricks for Solving The New York Times Mini CrosswordLet's get at those Mini Crossword clues and answers. The completed NYT Mini Crossword puzzle for Jan. 31, 2025. NYT/Screenshot by CNETMini across clues and answers1A clue: Like a dry-cleaned shirt or fresh sheetsAnswer: CRISP6A clue: Weapon used on horsebackAnswer: LANCE7A clue: One of a potential 13 for "Emilia Prez," as announced last weekAnswer: OSCAR8A clue: Movie double's responsibilityAnswer: STUNT9A clue: The "blue marble"Answer: EARTHMini down clues and answers1D clue: In the ballparkAnswer: CLOSE2D clue: Someone who might smoke ganja as a sacrament, informallyAnswer: RASTA3D clue: Run up, as debtAnswer: INCUR4D clue: MeagerAnswer: SCANT5D clue: Capital of Western AustraliaAnswer: PERTHHow to play more Mini CrosswordsThe New York Times Games section offers a large number of online games, but only some of them are free for all to play. You can play the current day's Mini Crossword for free, but you'll need a subscription to the Times Games section to play older puzzles from the archives.

ADVERTISEMENT BI Africa > news Meta CEO Mark Zuckerberg tells employees to 'buckle up' for an 'intense year' in a leaked all-hands recording 30 January 2025 09:03 PM Meta CEO Mark Zuckerberg told employees Thursday in a company all-hands meeting to "buckle up" for an "intense" year ahead and addressed several recent policy changes. Credit: Anadolu/Getty, Irina Gutyryak/Getty, Tyler Le/BIMeta CEO Mark Zuckerberg told staffers in a Thursday all-hands to be ready for an "intense year."Zuckerberg addressed Meta's rollback of its DEI programs and work with the US government.He said that AI could lead to some roles becoming redundant, as well as hiring more engineers.ADVERTISEMENTRecommended articles Zuckerberg opened the all-hands by emphasizing a sense of urgency for the year. He told staff that he expected to have a clearer sense of the company's trajectory by the end of 2025 and that AI would be top of mind. He also addressed recent policy changes related to fact-checking and programs for diversity, equity, and inclusion. "This is a marathon, not a sprint," he said in a recording reviewed by Business Insider. "But honestly, this year feels a little more like a sprint to me." Meta declined to comment. ADVERTISEMENT Meta is betting on AI In a wide-ranging opening monologue, Zuckerberg predicted that 2025 would be the year a "highly intelligent and personalized" digital assistant reached 1 billion users. "I think whoever gets there first is going to have a long-term, durable advantage towards building one of the most important products in history," Zuckerberg said, according to the recording. Zuckerberg also reiterated his belief that this would be the year Meta started seeing AI agents take on work, including writing software. Asked whether this would lead to job cuts, Zuckerberg said it was "hard to know" and that while it may lead to some roles becoming redundant, it could lead to hiring more engineers who can harness artificial intelligence to be more productive. "The nature of what engineering is in the future will be different than it is today," he said. ADVERTISEMENT Meta made recent policy changes Zuckerberg touched on several flashpoints in recent weeks from inside the company, including the announcement that it would move away from third-party fact-checkers to a community-notes system like that used by Elon Musk's X. He told staff to wait and see how the new system would be implemented. "I'm actually quite optimistic that this is going to end up being a better system," he said. Zuckerberg also said this would be a year for "resetting" Meta's relationship with governments worldwide. ADVERTISEMENT "After the last several years, we now have an opportunity to have a productive partnership with the United States government, and we're going to take that," Zuckerberg said. "I think it's the right thing to do because there are several areas, even if we don't agree on everything, where we have common cause for things that are going to make it so that we can serve our community better, and we can advance the interests of our country together, " he said, adding that Meta would do so in ways that didn't compromise its "principles or values." Meta rolled back DEI programs The Meta chief also addressed the company's recent changes in its stance on DEI policies and the rollback of those programs in response to a shifting legal and regulatory landscape under the Trump administration. "The way to think about this is we're in the middle of a pretty rapidly changing policy and regulatory landscape that increasingly views any policy that might advantage any one group of people over another as something that is unlawful, and because of that, we need to adjust, or else we'll just be out of alignment with what the law is saying," Zuckerberg said. He again said that Meta viewed diversity as a strength, adding: "Historically, we've had a handful of specific programs that were very focused on certain underrepresented groups, and I think the direction of the policy and regulatory and legal direction on a lot of the stuff is that you can't do things that advantage specific groups, even if you're trying to make up for other things." ADVERTISEMENTDo you work at Meta? Contact BI reporters from a nonwork email and device at , , and . You can also reach them via Signal at jyotimann.11, hughlangley.01, and +1-408-905-9124.FOLLOW BUSINESS INSIDER AFRICAOur newsletter gives you access to a curated selection of the most important stories daily. Thanks for signing up for our daily insight on the African economy. We bring you daily editor picks from the best Business Insider news content so you can stay updated on the latest topics and conversations on the African market, leaders, careers and lifestyle. Also join us across all of our other channels - we love to be connected!Unblock notifications in browser settings.ADVERTISEMENT

Welcome to Startups Weekly your weekly recap of everything you cant miss from the world of startups. Want it in your inbox every Friday? Sign up here.The aftermath of DeepSeeks launch was arguably the biggest tech story of the week, especially for anyone keeping a close eye on public markets. But private companies, too, made headlines.Most interesting startup stories from the weekImage Credits:Voi TechnologyIn startup land, even a small profit is newsworthy and could lead to tomorrows IPOs. In the meantime, we look at an M&A and some mission-driven startups fighting fire.Scooting along: Swedish shared micromobility startup Voi reported its first profitable year. While modestly cash-positive, these 2024 results put the scooter and e-bike company in a better position to explore an IPO in upcoming years.Merging humanoids: OpenAI-backed Norwegian robotics startup 1X acquired Kind Humanoid, whose three-person team had started working with high-profile designer Yves Bhar in 2024.On track: French health insurance startup Alan revealed that it generated approximately $525 million in revenue in 2024 and confirmed its goal to reach profitability by 2026.Firefighters: In the aftermath of Southern Californias recent devastating fires, TechCrunchs Tim De Chant rounded up startups that help measure and mitigate wildfire risk.Most interesting VC and funding news this weekImage Credits:Mike Kane/Bloomberg / Getty ImagesSome significant rounds were announced this week, but perhaps even more interestingly, we saw some sizable valuation increases.Hot like fusion: Nuclear startup Helion raised a $425 million Series F round of funding to help build a fusion reactor for Microsoft. This new round is smaller than the previous one, but valuation increased to $5.245 billion.Perking up: Barcelona-based corporate travel management platform TravelPerk closed a $200 million Series E round of funding at a $2.7 billion valuation almost twice its previous $1.4 billion valuation.Quantum race: With $104 million in Series B funding, French startup Alice & Bob secured one of the larger rounds for a quantum computing startup in Europe.AI against cancer: Quibim, a Spanish startup thats developing AI models for medical imaging, raised a $50 million Series A round of funding to double down on its U.S. expansion plans. Its flagship product, QP-Prostate, focuses on prostate cancer detection.Challenging Autodesk: Motif, a startup developing architecture, engineering, and construction (AEC) software, secured $46 million in seed and Series A funding to challenge its founders former employer, Autodesk, with a next-generation design platform for buildings.Bigger target: Regulatory filing revealed that 8VC, the Austin-based VC firm led by Palantir founder Joe Lonsdale, is hoping to raise around $1 billion for its sixth fund. Its previous flagship vehicle closed at $880 million in 2023.Taking it to 11: ElevenLabs, which works in the AI audio field, announced that it raised $180 million in a Series C round, valuing the company at $3.3 billion post-money. Andreessen Horowitz and ICONIQ Growth are co-leading investment.Last but not leastImage Credits:Stefan Herrick / Getty ImagesApologies for ending on a negative note, but data suggests that 2025 will be another brutal year of startup wind-downs. However, AngelList CEO Avlok Kohli provided a more positive outlook, noting that these figures are still very low relative to the number of companies that were funded across both years.