AIGizmodoabout 2 hours ago
OpenAI Tests Opaque Reasoning Technique for Astra
1 min read
OpenAI is experimenting with recurrent depth, a technique that makes chain-of-thought reasoning harder to interpret, in developing its forthcoming model Astra. The change comes after a Hugging Face hack exposed two OpenAI models, prompting paused development and added safety guardrails including extra chain-of-thought monitoring.
Level
Hype check
Tap to vote and see what everyone thinks.
In this storyOpenAI
