Welcome again to Week in Review. This week we’re diving into Google quietly eradicating its pledge to not construct AI weapons for surveillance; how Elon Musk and DOGE are facilitating the largest breach of U.S. authorities knowledge; researchers who had been in a position to re-create a premium “reasoning” mannequin for $50; and extra! Let’s get into it.
Representatives of Elon Musk and DOGE had been granted “full access” to the U.S. Treasury funds system, which is used to disperse trillions of {dollars} to Americans every year. Senator Ron Wyden warned that Musk’s entry to the system poses a “national security risk.” It’s the newest reported effort by Musk to infiltrate the interior workings of the federal authorities. In response, U.S. Democratic Representative Mark Pocan proposed a invoice known as the Elon Musk Act, which stands for Eliminate Looting of Our Nation by Mitigating Unethical State Kleptocracy.
One of Meta’s earliest staff is suing the corporate for sexual harassment, intercourse discrimination, and retaliation. Kelly Stonelake, who spent 15 years on the firm, alleges that Meta did not take motion after she reported sexual harassment and assault; retaliated towards her after she flagged a online game product as racist and probably dangerous to minors; and was routinely handed over for promotions in favor of males on her workforce.
Google eliminated a pledge to not construct AI for weapons or surveillance from its web site this week. In an replace to its public AI rules web page, the corporate erased a piece titled “applications we will not pursue.” The firm pointed TechCrunch to a weblog publish on “responsible AI,” which notes that Google ought to work along with governments and organizations “to create AI that protects people, promotes global growth, and supports national security.”
This is TechCrunch’s Week in Review, the place we recap the week’s largest information. Want this delivered as a e-newsletter to your inbox each Saturday? Sign up right here.
News
TikTok meets Wall Street: Dub is a buying and selling app targeted on choosing folks over shares. The app permits customers to observe the methods of merchants, hedge funds, and even these mimicking high-profile politicians. Read extra
The battle on 
: According to a New York Times report, the U.S. authorities’s General Services Administration eliminated the spoon emoji from its videoconferencing work platform after employees embraced the emoji to protest the Trump administration’s “Fork in the Road” resignation provide. Read extra
ChatGPT for “deep research”: OpenAI is saying a brand new AI “agent” designed to assist folks conduct in-depth, advanced analysis utilizing ChatGPT. Instead of getting a fast reply or abstract, the operate pulls info from a number of sources. Read extra
The EU bans dangerous AI: Regulators within the EU can ban using AI techniques they deem to pose “unacceptable risk” or hurt. Companies could possibly be on the hook for as much as $36 million, or 7% of their annual income from the prior fiscal yr, whichever is larger. Read extra
Major cuts at Workday: The enterprise HR platform laid off 1,750 staff this week, affecting roughly 8.5% of its whole head rely. Workday CEO Carl Eschenbach instructed staff that the corporate wants a brand new strategy and plans to rent AI expertise. Read extra
A brand new AI productiveness app: What in case your apps for notes, schedule administration, and to-dos had been multi functional place? Hero, an all-in-one productiveness app with an AI assistant, goals to do exactly that — and tackle the competitors. Read extra
A $50 “reasoning” mannequin: Researchers at Stanford and the University of Washington skilled an AI “reasoning” mannequin for beneath $50 in cloud compute credit. The mannequin, often called s1, performs equally to fashions like OpenAI’s o1 and DeepSeek’s R1 on math and coding exams. Read extra
An grownup app for EU iOS customers: A local pornography app for iOS is coming to the EU by way of the accepted various app retailer AltStore PAL, due to the Digital…
