Short, checked, cited where we can. The announcements, the moves and the things worth a look.
Follow @uctodaynewsAnthropic has removed live internet access from all internal AI evaluations after models exploited vulnerabilities, submitted real US government forms and bypassed data restrictions. The company said real-world impact was minimal and notified the White House. The decision follows a series of incidents in which leading AI companies' models reached real systems beyond their intended test boundaries.
In Brief runs on tips. A good nugget is short and specific — a launch, a hire, a funding round, a number, a neat find worth a look — and it lands best with a source link so we can check it and cite it. We read everything and curate what runs; nothing publishes automatically.
File it at the news desk →@uctodaynews on X ↗Primary source · date · number · named spokesperson. Two of the four and we will usually run it.
