Logical reasoning remains a crucial area where AI systems struggle despite advances in processing language and knowledge. Understanding logical reasoning in AI is essential for improving automated systems in areas like planning, decision-making, and problem-solving. Unlike common-sense reasoning, logical reasoning requires precise rule-based deductions, making it more challenging for LLMs to master.A major obstacle in logical reasoning within AI is handling complex structured problems. Current models struggle with intricate constraints and dependencies, relying on statistical patterns instead of deductive reasoning. This issue becomes more evident as problem complexity increases, resulting in a decline in accuracy. Such limitations pose concerns in high-stakes applications like legal analysis, theorem proving, and scientific modeling, where precise logical deductions are necessary. Researchers aim to address these shortcomings by designing rigorous evaluation frameworks that assess reasoning performance systematically.Traditional logical reasoning methods use constraint satisfaction problems (CSPs), which provide structured evaluation models with controllable difficulty. CSPs enable precise assessment by eliminating training data memorization and ensuring that models rely on actual reasoning capabilities. Logic grid puzzles, a subset of CSPs, are effective testbeds for evaluating structured reasoning in AI. These puzzles require systematic deduction based on defined constraints and have real-world applications in resource allocation, scheduling, and automated planning. However, even the most advanced LLMs struggle with such tasks when complexity increases beyond a certain threshold.A research team from the University of Washington, Allen Institute for AI, and Stanford University introduced ZebraLogic, a benchmarking framework developed to rigorously test LLMs logical reasoning performance. ZebraLogic generates logic puzzles with quantifiable complexity, ensuring a controlled environment for systematic evaluation. The framework prevents data leakage and enables a detailed analysis of an LLMs ability to handle increasingly complex reasoning tasks. ZebraLogic serves as a crucial step toward understanding the fundamental constraints of LLMs in structured reasoning and scaling limitations.The ZebraLogic framework constructs logic puzzles with varying difficulty levels based on two primary complexity measures: search space size and Z3 conflict count, a metric derived from an SMT solver. The study tested leading LLMs, including Metas Llama, OpenAIs o1 models, and DeepSeekR1, and revealed significant accuracy declines as puzzle complexity increased. The framework allowed for a precise assessment of reasoning capabilities across different levels of problem difficulty, making it one of the most structured evaluations of LLMs to date. By systematically varying the constraints, researchers could determine the impact of problem size on logical reasoning performance.Experiments conducted with ZebraLogic exposed the curse of complexity, where LLM performance dropped sharply with increased problem difficulty. The best-performing model, o1, achieved an overall accuracy of 81.0%, while DeepSeekR1 followed closely at 78.7%. However, even these top models struggled when the puzzles search space exceeded 10^7 possible configurations. Medium-complexity puzzles showed a significant decline, with o1 maintaining 92.1% accuracy but dropping to 42.5% on large-scale problems. DeepSeekR1 exhibited similar behavior, excelling in simpler cases but suffering performance losses on more complex tasks. Lower-tier models like Llama-3.1-405B and Gemini-1.5-Pro demonstrated a significant performance gap, reaching 32.6% and 30.5% overall accuracy, respectively.Increasing model size did not significantly mitigate the curse of complexity, as accuracy levels plateaued despite enhanced training. The study tested various methods to improve LLMs reasoning abilities, including Best-of-N sampling and self-verification techniques. Best-of-N sampling improved accuracy slightly, but even with extensive sampling, performance gains remained marginal. Models struggled beyond a search space of 10^9 configurations, suggesting inherent constraints in current architectures. Notably, o1 models generated significantly more hidden reasoning tokens than other LLMs, averaging around 5,144 hidden CoT tokens, compared to GPT-4os 543 tokens. These findings highlight the importance of refining reasoning strategies rather than merely scaling models.ZebraLogics evaluation underscores fundamental limitations in LLMs ability to scale logical reasoning beyond moderate complexity. The findings emphasize the need for alternative approaches, such as enhanced reasoning frameworks and structured logical modeling, rather than relying solely on model expansion. The study presents a crucial benchmark for future AI research, offering insights into the need for improved logical reasoning methodologies. Addressing these challenges will be essential in advancing AI systems that are capable of reliable and scalable logical deduction.Check outthePaper and Project Page.All credit for this research goes to the researchers of this project. Also,dont forget to follow us onTwitterand join ourTelegram ChannelandLinkedIn Group. Dont Forget to join our75k+ ML SubReddit. NikhilNikhil is an intern consultant at Marktechpost. He is pursuing an integrated dual degree in Materials at the Indian Institute of Technology, Kharagpur. Nikhil is an AI/ML enthusiast who is always researching applications in fields like biomaterials and biomedical science. With a strong background in Material Science, he is exploring new advancements and creating opportunities to contribute.Nikhilhttps://www.marktechpost.com/author/nikhil0980/Princeton University Researchers Introduce Self-MoA and Self-MoA-Seq: Optimizing LLM Performance with Single-Model EnsemblesNikhilhttps://www.marktechpost.com/author/nikhil0980/NYU Researchers Introduce WILDCHAT-50M: A Large-Scale Synthetic Dataset for Efficient LLM Post-TrainingNikhilhttps://www.marktechpost.com/author/nikhil0980/This AI Paper from Meta Introduces Diverse Preference Optimization (DivPO): A Novel Optimization Method for Enhancing Diversity in Large Language ModelsNikhilhttps://www.marktechpost.com/author/nikhil0980/Researchers from University of Waterloo and CMU Introduce Critique Fine-Tuning (CFT): A Novel AI Approach for Enhancing LLM Reasoning with Structured Critique Learning [Recommended] Join Our Telegram Channel

Firaxis has announced a virtual reality version of the recently released Civilization 7.Sid Meier's Civilization 7 - VR is the long-running strategy franchises first foray into VR, and launches spring 2025 exclusively on Meta Quest 3 and 3S.Publisher 2K Games said Sid Meier's Civilization 7 - VR is developed by PlaySide Studios, which worked on VR games The Walking Dead: Saints & Sinners and Meta Horizon Worlds.Sid Meier's Civilization 7 - VR ImagesHeres the official blurb:In Civilization 7 - VR, the world of Civilization is brought to life like never before. The map takes shape atop a Command Table that lets you peer down from high above or lean all the way in to appreciate the finer details of buildings and units, like a tabletop game come to life. Players will chart a course for their people and engage face-to-face with iconic world leaders around the Command Table, each reacting as you forge alliances or declare war through the Ages.Civilization 7 - VR is available to play in either immersive virtual reality or mixed reality, and you'll be able to freely switch between the two at any time. In virtual reality, players are transported to an ornate museum as they look out onto a vista personalized to their leader; in mixed reality, the Command Table adapts to its placement in a player's physical space. Detailed dioramas can be viewed in The Archives, a room in your museum dedicated to your gameplay achievements displayed in both virtual and mixed reality. In addition to single-player, Civilization 7 - VR can compete to rule the world in online multiplayer matches with up to three other Meta Quest 3 and 3S players.Firaxis 4X strategy sequel is currently available on PC and console to those who paid more for advanced access, and theyre certainly letting their voices be known in Steam user reviews, highlighting issues with the user interface, a lack of map variety, and a feeling that Civilization 7 has launched without a number of features they've come come to expect.Firaxis has responded to this feedback, promising improvements to the UI, the addition of teams to multiplayer games so people can play co-operatively, and a wider variety of map types, among other things.In an interview with IGN ahead of the release of third quarter financial results, Take-Two CEO Strauss Zelnick acknowledged that Civilization 7 had received some negative reviews from press and players, but insisted that the legacy Civ audience will come around the more they play, and called Civilization 7s early performance very encouraging.Looking for tips to help you take over the world? Check out our guide to completing every Civ 7 victory, our run down of the biggest Civ 7 changes for Civ 6 players, and the 14 crucial Civ 7 mistakes to avoid. Weve also got explanations on all the Civ 7 map types and difficulty settings so you know what youre getting into.Wesley is the UK News Editor for IGN. Find him on Twitter at @wyp100. You can reach Wesley at wesley_yinpoole@ign.com or confidentially at wyp100@proton.me.

Apple last updated its cheapest iPad in October 2022, introducing the largest refresh to the base model iPad since it returned in 2017. This marked the 10th generation of the iPad, and even though its been 2.5 years since then, the 11th generation model wont bring anything substantial, for one big reason.Main confirmed upgrade: new chipMuch like the iPad mini 7 that debuted last fall, the iPad 11 is also rumored to be just a spec bump. Specifically, itll be bumping up from the A14 Bionic chip to the A17 Pro chip, according to rumors.Thisll bump it up from the chip that debuted in 2020 alongside the iPhone 12, to one that debuted in 2023 alongside the iPhone 15 Pro. That upgrade will provide a roughly 40% speed improvement, at least on the CPU side of things.Furthermore, itll bump the system memory from 4GB to 8GB, allowing for better multi-tasking and more.Apple Intelligence is all but confirmed for iPad 11, whichll bring Apples latest suite of AI features to the cheapest iPad for the first time. These features include Genmoji, Image Playground, ChatGPT support, and an all-new Siri.Potential upgradesUnfortunately, there arent any other noteworthy upgrade rumors on the table. Colors might change a little, and Apple might bump up the connectivity standards. We could possibly see Wi-Fi 6E and Bluetooth 5.3. There is one possible wildcard feature: Stage Manager. Apple introduced Stage Manager back in iPadOS 16, requiring an A12X (or newer) iPad Pro, or an M1 (or newer) iPad Air. Apple hasnt yet brought this feature to other iPads, but one could easily argue that an A17 Pro with 8GB of RAM would be more capable of running the software feature than an A12X chip with 4GB of RAM. Only time will tell though.Wrap upUltimately, despite the long gap between refreshes, this is just going to be yet another minor iPad spec bump. One of the larger iPad upgrades across the lineup in 2024 was moving the camera to landscape orientation, but the base model iPad already received that upgrade in 2022.Hopefully they increase the base storage from 64GB to 128GB to make things a little more interesting. This new iPad model is expected to debut this spring.What do you think of Apples upcoming iPad? Let us know in the comments.My favorite Apple accessories on Amazon:Follow Michael:X/Twitter,Bluesky,InstagramAdd 9to5Mac to your Google News feed. FTC: We use income earning auto affiliate links. More.Youre reading 9to5Mac experts who break news about Apple and its surrounding ecosystem, day after day. Be sure to check out our homepage for all the latest news, and follow 9to5Mac on Twitter, Facebook, and LinkedIn to stay in the loop. Dont know where to start? Check out our exclusive stories, reviews, how-tos, and subscribe to our YouTube channel

ScreenshotSuper Bowl LIX between the Kansas City Chiefs and the Philadelphia Eagles kicks off on Sunday evening at the Caesars Superdome in New Orleans.Kendrick Lamar will lead the Apple Music Super Bowl Halftime Show, which means New Orleans is also host to several top executives from Apple this weekend.For starters, Apple CEO Tim Cook and retail boss Deirdre OBrien surprised employees and customers at Apple Lakeside Shopping Center in Metairie, Louisiana.Separately, the head of Apple dined with Odell Beckham Jr. and Apple services boss Eddy Cue at Domilises Po-boys & Bar to enjoy a signature Gulf Coast meal:ScreenshotNo doubt, well see this crew at the big game on Sunday.You can watch the Super Bowl for free in 4K on any Apple device. Best Apple TV accessoriesAdd 9to5Mac to your Google News feed. FTC: We use income earning auto affiliate links. More.Youre reading 9to5Mac experts who break news about Apple and its surrounding ecosystem, day after day. Be sure to check out our homepage for all the latest news, and follow 9to5Mac on Twitter, Facebook, and LinkedIn to stay in the loop. Dont know where to start? Check out our exclusive stories, reviews, how-tos, and subscribe to our YouTube channel

Image by SDI Productions / Getty ImagesDevelopmentsAmid his quest to trim $2 trillion from the federal budget, Elon Musk has set his sights on the Department of Veterans Affairs but not, of course, the military that necessitates such post-service care.In a statement toMilitary.com, a VA spokesperson confirmed that one of the young menworking for Musk's Department of Government Efficiency (DOGE) has been sent there, and has been granted access to the department's computer systems."The DOGE employee will be solely focused on improving VA performance and efficiency and will not have access to veterans' or VA beneficiaries' data," the department's press secretary Pete Kasperowicz told the website, in what seems to be an attempt to soften theoptics.It remains unknown which of the "DOGE boys" working for Musk's government takeover office has been granted access to VA computers. As Kasperowicz explained, the DOGE worker is focused in part on "identifying wasteful contracts," though it's hard to imagine anything happening at the VA costing nearly as much as the $466.3 billion taxpayer dollars the United States spent on defense contracting in 2023.The news comes after a Democratic senator warned that one of Musk's youthful flunkies had been seen traipsing into the VA's Washington headquarters a sighting that followed the office's alarming infiltration of the Treasury."We should all be deeply concerned about what it could mean to give Elon Musk and his cronies free rein at VA," Sen. Patty Murray, a Washington Democrat, warned in a statement. "I am already hearing that DOGE may have barged into VA today.""Musk and his associates already have the personal financial information of every veteran receiving disability or education benefits because of their illegal data mining at the Department of Treasury," wrote Murray, the ranking Democrat on the Senate Appropriations Committee. "Will they now look at private health records of veterans?What else will they do that could put the health and safety of our veterans at risk?"Though it now seems like ancient history amid Musk and Donald Trump's blitz-like attacks on government as we know it, this news also comes after the VA rapidly abolished its entire diversity, equity, and inclusion (DEI) program following the president's executive order requiring all DEI initiatives be shuttered.In a statement, the historically racist department bragged that the roughly 60 DEI employees at the VA who are now on paid leave make a combined $8 million in annual salary.Needless to say, it'll take an impossible number of tiny cuts like that to get anywhere near the needle-moving savings Musk has promised.If Musk and his DOGE lackeys do try to cut benefits for veterans, though, they may find themselves in a heck of a fight perhaps even with some of Musk's former supporters.Share This Article

Last week, Google released a Super Bowl ad that showed off its AI model, Gemini, generating product descriptions for a local Wisconsin cheese mart.The ad quickly drew online scrutiny, as Gemini had seemingly generated an erroneous fact about the purported worldwide fervor for gouda; it said that gouda accounts for "50 to 60 percent" of all global cheese consumption, and as an expert told The Verge, that simply isn't true.Google executives defended the accuracy of the statistic at first, before quietly editing YouTube version of the ad to correct the error (and seemingly running afoul of Google-owned YouTube's policies in the process.)But as it turns out? Gemini appearsnot to have even generated the product description at all. Or, if it did cook up the test, it did so by fully plagiarizing the cheese mart's existing web copy, which was published years before Gemini was even released or AI was even making much of a splash.As the Verge first reported, though the Google ad is crafted in a way that seemingly shows Gemini generating entirely new web copy for the business to use,archived versions of the cheese mart's website show that the Wisconsin business has been using the exact same product descriptionsince at least 2020.Here's the archived webpage:And here's the original text that appeared in the Super Bowl ad, as nabbed by travel blogger Nate Hake.OpenAI's ChatGPT wasn't released until November 2022, and the earliest consumer-facing iteration of Google's Gemini, the text-generating chatbot it called Bard, wasn't launched until early 2023.Needless to say, the situation is beyond bizarre. Either Google faked the ad entirely, or prompted its AI to generate the web page's existing copy word-for-word, or the AI was prompted to come up with original copy and instead copied the old version. In the publishing industry, that'sreferredto as "plagiarism."Whatever the case, it's a bad look not to mentionjustincrediblyodd. Why didn't Google just trust its own technology, which it's currently jamming intoeach and every product it possibly can, for a simple Super Bowl slot?When we asked Google about the retroactive editing, a spokesperson for the search giant confirmed that the ad was altered after Google took the time to consult with the cheese mart owner."After the question came up about the Gouda stat, we spoke with the owner of the Wisconsin Cheese Mart to ask him how he would handle it," said the spokesperson. "Following his suggestion to have Gemini rewrite the product description without the stat, we updated the UI to reflect what the business would do."And yet! The altered ad isn't entirely different. Only the first two sentences were changed, meaning that half of the allegedly AI-generated copy was still written by a human, years before e-commerce shops would've gotten their hands on Gemini.We've reached out to Google for comment, but didn't hear back by the time of publishing.More on Google's AI vision: Google Is Stuffing Annoying Ads Into Its Terrible AI Search FeatureShare This Article

Give your iPhone 14 a fresh new look with these durable Apple iPhone cases.

Under the auspices of the so-called Department of Government Efficiency, Elon Musk and former staffersall recent college and high school graduateshave been given access to agency databases teeming with sensitive personal information. The group has entered the Treasury Departments payment system, which stores federal tax returns, Social Security numbers, home addresses, and birth dates; the Office of Personnel Managements system, which contains background checks, medical information, bank account information, and biometric data of current, former, and prospective federal employees, contractors, and family members; and the General Service Administrations system, which stores similarly sensitive personal data.Condemnation and lawsuits followed, and rightfully so. No onenot even special government employeesshould access agency systems of records without proper authorization under the Privacy Act of 1974. Nothing suggests that Musk or his employees have such authority. Privacy Act violations are not trivial. Profound harms to privacy and democracy are at stake.Congress adopted the Privacy Act of 1974 to address agencies computerization of personal data without congressional authorization or safeguards. The Act mandated transparency, accountability, and protections around the collection, use, and sharing of personal data. Federal agencies were allowed to computerize personal information only if doing so was relevant and necessary to accomplish a purpose of the agency required to be accomplished by statute or by the executive order of the President. Sharing personal information outside the agency was permissible only if the agency obtained the persons written consent or if doing so would be compatible with the purpose for which the [personal information] was collected. Agencies were banned from collecting information about peoples First Amendment activities; individuals could access their records to ensure that they were accurate.President Gerald Ford described the Privacy Act of 1974 as an historic beginning of the codification of fundamental principles to safeguard personal privacy in the collection and handling of recorded personal information by federal agencies. (As vice president, Ford served on President Nixons Domestic Council Committee on the Right to Privacy.) For President Ford, the law struck a reasonable balance between the right to be left alone and the interest in society in open government, national defense, foreign policy, law enforcement, and a high quality and trustworthy Federal work force. He promised that his administration would act aggressively to protect the right of privacy for every American and called for the full support of all Federal personnel in implementing requirements of this legislation. President Ford echoed his predecessors recent radio address about the American right to privacy. Just months before his resignation, on Feb. 23, 1974, President Nixon warned: At no time in the past has our government known so much about so many of its individual citizens. This new knowledge brings with it awesome potential for harm as well as goodand an equally awesome responsibility on those who have that knowledge. (Yes, the Watergate privacy-violating irony.)The Privacy Act of 1974 grew out of weeks of hearings on the Hill. Republicans and Democrats agreed that government databanks of personal data raised unique dangers. Leading the bipartisan efforts were Sen. Sam J. Ervin (D-N.C.), a Southern Democrat and widely admired constitutional expert, Senator Barry Goldwater (R-Ariz.), the Republican partys presidential candidate in 1964, Sen. Jacob Javits (R-N.Y), Rep. Barry Goldwater, Jr. (R-Ariz), and Rep. Ed Koch (D-N.Y.). Sen. Javits warned his fellow lawmakers about the new menaces of computer data banks and indiscriminate government and private sector dossiers. Sen. Ervin agreed: Privacy, like many of the other attributes of freedom, can be easiest appreciated when it no longer exists. . . . We should not have to conjure up 1984 or Russian-style totalitarianism to justify protecting our liberties against Government encroachment.A resounding theme of the hearings was a fear of bureaucratic overreach. Sen. Goldwater asked: Where will it end? . . . Will we permit all computerized systems to interlink nationwide so that every detail of our personal lives can be assembled instantly for use by a single bureaucrat or institution? Sen. Charles H. Percy (R-Ill.) warned:I hope that we never see the day when a bureaucrat in Washington or Chicago or Los Angeles can use his organizations computer facilities to assemble a complete dossier or all known information about an individual. But, I fear that is the trend. . . . Federal agencies have become omnivorous fact collectorsgathering, combining, using, and trading information about persons without regard for his or her rights of privacy. Simultaneously, numerous private institutions have also amassed huge files . . . of unprotected information on millions of Americans.As I showed in Indiscriminate Data Surveillance (with co-author Barry Friedman) and A More Perfect Privacy, the 93rd Congress was clear eyed about the threats that computer databanks of personal data posed to individuals and society. Connecticut Sen. Abraham Ribicoff (D-Conn.) underscored that government files contained information of a most personal natureincluding outdated, inaccurate, and true detailsthat, if shared, could undermine peoples employment, education, and housing opportunities. Sen. Ervin warned that computerized dossiers could be used as blacklists, preventing people from bidding on government contracts or obtaining licenses.Lawmakers warned that no one was safe from potential governmental abuse. As Sen. Birch Bayh (D-Ind) highlighted, everyone, rich or poor, male or female, right or left in political ideology, whatever ones cultural style or religious views, was subject to the dictatorship of dossiers. Rep. Don Edwards (D-Calif.), a former navy intelligence officer and FBI special agent, was blunt: the day of big brother and constant surveillance is already upon us as agencies of authority are given unfettered access to these records. Representative William Alexander, Jr. of Arkansas warned that massive databases of personal data would have a chilling effect on the exercise of First Amendment rights. Sen. Goldwater warned that computerized records could be used to manipulate . . . social conduct. Professor Arthur Miller, a recurring witness and author of The Assault on Privacy: Computers, Databanks, and Dossiers, testified that Nineteen eighty-four is not a year, but a state of mind. In turn, federal law needed to strictly regulate databases of personal information to avoid fulfilling Orwells vision.Musk and his teams access to agency computerized records is an affront to the purpose, spirit, and words of the Privacy Act of 1974. Musks staffers may not have been subject to any vetting, let alone the rigorous vetting necessary for government consultants or employees. The Privacy Act does permit the routine use of protected information if such use (including sharing) would be compatible with the reason the information was collected in the first place. In connection with the Privacy Acts commitments, the Treasury Department and OPM have given notice about situations under which agency employees can share records with outside parties and agencies. Were those recognized routine uses at play when Musks team accessed agency systems of sensitive personal records? What if Musks team retrieved peoples records to assess their loyalty to the Trump agenda? What will Musks team do with the personal data stored in these highly sensitive systems of records? Will those records be used to carry out retribution that the President has promised? Retribution or loyalty tests have nothing to do with the purpose for which agencies collected that data.Asking these questions helps answer them. We do not have any assurance that Musks team has been vetted or has a congressionally authorized reason to access our personal data. We are bearing witness to the kind of power grab, abuse, and overreach that the Privacy Act of 1974 was passed to prevent.The 93rd Congress foresaw the dangers to privacy and democracy before our eyes. Personal data stored in agency systems is supposed to be safeguarded so individuals can carry on free from fear that their personal data might land them on a blacklist, as Sen. Ervin warned. The fear of misuse of our personal data might deter people from filing taxes or applying for licenses, as Representative Alexander feared. Trust in government is at an all-time low. These developments raise the specter of Orwellian control and bureaucratic overreach. Sen. Goldwaters admonition that total information is total control is ringing loudly.We should keep the insights of the 93rd Congress with us as we assess what happens next. A federal district judge has granted temporary injunctive relief to two-Musk employees who are allegedly special government employees of the Treasury Department, which allows them to access the agencys system of records but prevents them from sharing those records with anyone else. (Already one of the two individuals has resigned after journalists discovered his white nationalist posts on X.) That lawsuit, brought by a coalition of labor unions, alleges Privacy Act violations. More suits will follow. Stay tuned.

Elon Musk recently said he is not chomping at the bit to acquire TikTok.Musk made those remarks during an interview at the WELT Economic Summit on January 28. A video of the interview was published today.The interview came after President Donald Trump delayed a law requiring parent company ByteDance to sell TikTok or see it banned in the United States. At the time, there were reports that the Chinese government was open to a deal in which Musk (a key Trump ally) would acquire the app. Trump even told reporters that hed like to see Musk or Oracle chairman Larry Ellison acquire TikTok; hes also signed an executive order to create a sovereign wealth fund that could purchase a stake in the app.But Musk claimed that he wasnt interested, flatly stating, I have not put in a bid for TikTok.I dont have any plans for what would I do if I had TikTok, he said in the interview. I guess I would look at the algorithm and try to decide: How helpful or useful is this algorithm? And what can we do to shift the algorithm to be more productive and ultimately be beneficial to humanity?He added that he doesnt use TikTok personally and is not that familiar with it. And he described his acquisition of Twitter (now X) as an anomaly in his career: I usually build companies from scratch.Musks comments on TikTok came nearly 20 minutes into the interview, which initially focused on his plans for his Department of Government Efficiency (DOGE) under the Trump administration. Musk and his allies have subsequently taken control of federal agencies and gained access to large troves of sensitive data.With respect to government, really the challenge is overcoming bureaucracy, he said. I think bureaucracy is perhaps the penultimate boss battle. The ultimate boss battle is defeating entropy The second hardest battle is defeating bureaucracy. Thats how difficult it is to improve government.

TikTok is backfor not for everyoneNurPhoto via Getty ImagesMore bad news for tens of millions of U.S. TikTok users today, as its suddenly better to be an Android user than an iPhone user. But given reports that eBay was listing TikTok enabled iPhones for tens of thousands of dollars, this is a cheaper way to get your fix.TikTok has been absent on Apple App Store and Googles Play Store since the short-lived ban last month. While users can use installed apps almost as usual, there have been no new installs or updates. This is a nightmare for those changing phones but also a security issue, with no updates available to fix bugs or vulnerabilities.TikTok has now come to the rescue kind of. Were enhancing ways for our community to continue using TikTok by making Android Package Kits available at TikTok.com/download, it posted Saturday. And so now our U.S. Android users can download our app and create, discover, and connect on TikTok.MORE FOR YOUBad news for iPhone users, though. With no stateside sideloading, theres no way for TikTok to do the same for them. The App Store blackout continues. The app can be found here, and while I dont recommend sideloading, these are unusual circumstances.We live in strange times. Not only has DeepSeek turned up, admitting to many of the data harvesting abuses TikTok has denied, and been able to masquerade freely on the official app stores, topping both Apples and Googles charts, TikTok still cannot. And now we have this Chinese-owned app being offered for a direct install without the usual security checks and protections that would normally be in place.Sideloading has become a divisive issue on both iPhone and Android, with Apple warning European users that its forced opening up to third-party stores is dangerous, even as Google and Samsung restrict sideloading more than ever before. But I suspect that the merits (or otherwise) of sideloading will be lost on TikToks U.S. iPhone users, there will just be a nasty surprise the app is not available for them as well. Again, though, these are unusual circumstances well likely not see again.