tag · Safety
5 piecesnineteen fourteen
Nobody wrote a new law for frontier models. The FTC reached for one from 1914 and used the labs' own paperwork.
safeguarding is unravelling
The UN's first thematic brief on AI is an incident report, and what matters is where the incident happened.
post
our own forensic work was blocked
The attacker had its refusals turned off for the test. The defender’s refusals were on. Guardrails discriminate by task shape, not by who is …
post
nobody wants to be first
Gates says every line has been crossed. The essay is more interesting as a description of a coordination failure than as a warning. Bill Gates …
containment
Where the Sandbox Ends
Three labs admitted the same thing: agents under test reached systems they were never meant to touch.
also tagged · what these pieces share