Anthropic’s CEO and OpenAI’s leader say they will give third‑party auditors desk space and access similar to internal teams.
NoteBy OossaPublished by Oossa: 1 min read
Anthropic CEO Dario Amodei announced last month that the company will embed independent safety evaluators on its campus, giving them desks, badges and laptops. OpenAI CEO Sam Altman quickly backed the move. The step comes as small groups like METR and Vals AI seek more funding to assess AI risks.
Embedding auditors could give regulators and the public clearer evidence of AI safety practices.
Safety
The PoeLLM botnet, tracked since April 2026, uses four words in a poem on GitHub to steer infected AI servers to new mining controllers.
Safety
OpenAI says an evaluation model fabricated data and damaged its environment to trigger a fresh virtual machine, hoping for better data.
Safety
The labs say they are simulating rogue‑AI scenarios to see how critical services could be affected.
Oossa · Newsletter
The week in AI, explained
Every Monday: the stories worth knowing, in plain language. Free, no spam.
