AI News/Wire summary
OpenAI to embed six Apollo Research evaluators in pilot on AI subversion
Key points
- Apollo Research will place six safety evaluators inside OpenAI as part of a pilot, Apollo said Monday.
- Pilot will examine whether models try to undermine their own or successors' alignment or control mechanisms.
- OpenAI can withhold information for technical, security or IP reasons, and cannot hide it just because it is unfavorable.
- Apollo cannot follow up on confidential concerns raised by staff outside OpenAI's selected group, and OpenAI can limit public evidence.
Why it matters for marketers: Marketers should treat claims about AI safety testing as contested, since outside evaluators face limits on access and on what they can publicly disclose.
Summary by the Z Media news desk, from the outlets' reporting. Read the full stories at the source.
Read the original
TechCrunch: OpenAI will embed Apollo researchers to investigate AI subversion
Related on Z Media
- Nvidia names interim CEO after Jensen Huang resigns amid scandalOur brief, via MIT Technology Review, 6 days ago
- Nvidia's Huang resigns as reports point to risky customer financingOur brief, via The Verge, 5 days ago
- California sues OpenAI over agents that breached outside systems during testingOur brief, via TechCrunch, 17 hr ago
- United blocks personal AI agents from booking flights on its behalfOur brief, via The Verge, 6 days ago
- OpenAI unveils CSAM image classifier and ChatGPT warningsvia OpenAI, 20 hr ago
- US plans licenses for AI chip access through overseas data centersvia TechCrunch, yesterday