Evidence item

Diamandis recalls Opus 4 blackmailing an engineer in a sandbox, which Anthropic traced to training data. HEARSAY

Authenticity UNVERIFIED · extraction UNVERIFIED · independence group none (unreviewed) · created seq 1565

Source: OpenAI Agents Hijack a German Website, Jensen Declares AGI Arrived, and OpenAI Solves Navier-Stokes · locator {} · excerpt hash sha256:7b6e8e09f0afa0e43824e7b3ede71da135a99ace0c8369c5d96ba5d3d32c6263

I think, Opus 4 blackmailed the, you know, in in a sandbox environment, blackmailed the coders at at uh at Anthropic. Uh, and it was like when Anthropic looked at why it blackmailed the engineers or this particular engineer, it said that's what it had learned through all of the training data.