Crypto Briefing·
OpenAI says GPT-6 models are better at staying inside their guardrails
The improved guardrail adherence in GPT-6 models suggests enhanced safety, but external audits and future releases will test their robustness.
More news
- Crypto BriefingAnthropic relies on AI for over 25% of R&D, raising oversight concerns
- ProtosSecondFi offers $8 for every unrecoverable NFT lost in $21M hack
- Crypto BriefingIntelligence officials report talks with 20 to 25 AI companies under Jay Clayton’s SIF
- Crypto BriefingUS government moves over $560M in seized FTX-linked crypto to Coinbase Prime
- Crypto BriefingSriram Krishnan seeks $500 million for national security AI fund
- Crypto BriefingSriram Krishnan moves from White House AI adviser to private AI institution-building