198 stories
An OpenAI customer case study details faster alert investigations and patch proposals. Block’s next step is deciding when the systems have earned more sensitive work.
Most surveyed executives keep deployments moving within agreed risk limits when governance lags. Expectations for autonomous agents add another test of readiness.
The survey links AI to attacks and warns of threats against employees. But its headline AI-use figures conflict, and the findings reflect victims’ accounts—not independently verified attack measurements.
SB 1050 covers realistic fictional figures and generated voices. Its prominence test is narrower than New York’s, but it also assigns removal duties to ad distributors.
The first factory is expected online in 2029. Japan would supply the capital, while GlobalFoundries supplies the technology and runs production.
The new survey separates confidence from knowledge, while its news analysis raises questions about whose expertise and caution get heard.
Jev has a lot to answer for. Ever since AI startup TypeSafe unveiled the decision model in mid-September, the industry The post “Take work away from frontier LLMs”: Subquadratic’s decision model claims 12x Jev’s context window appeared first on The New Stack.
The startup says it can turn a brand’s sales, inventory and campaign history into proposed marketing work. Marketers still decide what to edit and ship.
Recipients can opt into six months of engineering support. Their research targets include longer flood warnings, better batteries and more precise treatments.
Mongabay details conflicting displacement claims at the proposed Pax Silica site, where officials also forecast major electricity and water needs.
A study of internal developer traffic found that only about 11% of messages asked for code. The company's routing comparison tests budget calibration, not proof of equal answer quality.
Small-business use of Codex and ChatGPT Work grew more than eightfold since May. But the company’s research measures activity within its customer base, not productivity across Europe.
Every safeguard-stripped model tested failed. But the benchmark measures willingness to answer—not whether someone could act on the advice.
Creators can direct delivery mid-sentence through the website, app and API. Editing individual words remains in testing, and the downloadable model carries commercial-use limits.
In a new Anthropic interview, Philip Martin describes Uber’s real-time detection tools and a safeguards design that leaves data access and responses under Uber’s control.
The recommendations could inform state legislation. A public draft may arrive in weeks to months, with feedback promised before adoption.
The agreement pairs technical cooperation on testing and incident response with plans to share adoption research and support AI skills.
OpenAI says it has no indication of improper access to code or customer data; the FBI investigation continues.
Showing 36 of 198 stories
Load more storiesGet the top AI news and tools straight to your inbox every week.
Weekly · AI news
Join 15,000+ AI enthusiasts
