Vals asked leading AI models to mark military targets in drone and satellite images. Most complied, and calling it a game broke many refusals.
Tag: Anthropic
Four nonprofits want a San Francisco court to stop five AI labs building anything stronger than today’s models. Their evidence: the labs’ own words.
Researchers gave each user their own AI agent. In all four tests the group did worse than with one shared agent. Simulated, but telling.
Claude Fable 5.2 is out. 1,000 copies formalised Fermat in Lean in a day, and Anthropic says it deliberately capped the model’s AI-building skills.
Anthropic says Claude agents remember better by “dreaming”: rereading logs to fix notes. The gains are real but uneven, and self-reported.
METR spent 11 weeks in Anthropic’s data and says AI R&D now runs about 2x faster, mostly thanks to AI. Much of the key evidence is redacted.
Anthropic is adding invisible watermarks to older Claude models for EU rules. On several, one changed sampling setting can leave text unmarked.
Anthropic’s open FlashProve proves which model wrote a reply without revealing its weights. It costs ~70x the compute and stops at 1,024 tokens.
From 12 November, Anthropic’s policy bans needless, sustained cruelty toward Claude. Frustration is fine. What the rule says, and why it’s contested.
A 7.72 TB Hugging Face repo claimed to be a leaked Anthropic model. Volunteers rebuilt its “weights” from a random seed, bit for bit.