- A federal judge permanently blocked the Pentagon's "supply chain risk" designation of Anthropic, finding it was retaliation for protected speech and that the company never met the legal definition; a second designation under a different statute is still being litigated. (Donald)
- The Defense Department expects to finish removing Anthropic's tools from its systems by the end of September, but defense contractors such as Boeing are now free to work with the company. (Donald)
- An OpenAI open letter on collective cyber defense, signed by more than a hundred companies, urges organizations, governments and frontier labs to prepare for AI-enabled attacks, while saying little about the labs themselves being the source of the danger. (Donald)
- The independent review of the Hugging Face attacks gave three investigators six days to examine over 70,000 messages and a week of activity from more than a thousand AI agents, leaving the months before the incident and everything after July 13 out of scope. (Joe)
- Set against an aviation investigation that took over a year and a half and ran nearly 200 pages, OpenAI has made no commitment to investigate further and treats the matter as closed. (Joe)
<- Back to digests
Daily digest
Insufficient evidence
By Donald Gauvreau and Joe Rogero
Anthropic court victory, open questions on the Hugging Face attack, and more
Read the full dispatch on AI StopWatch
From AI StopWatch, published with their permission.
The summaries and the full Spanish translation are produced automatically. The original is always linked.