ainews.cx Entities

Hacker News OpenAI

Every Reward Bends

How to run an internal security incentive program without it getting gamed, and what OpenAI's Hugging Face incident shows about rewards bending under pressure.

Topics: Models

Entities: OpenAIModels

Hacker News OpenAI + 1 source (Hacker News OpenAI and Google News OpenAI Models)

wsj.com

Article URL: https://www.wsj.com/tech/ai/openai-to-restrict-astra-model-after-rating-it-critical-cyber-risk-499b5a46 Comments URL: https://news.ycombinator.com/item?id=49528347 Points: 6 # Comments: 4

Topics: Models

Entities: OpenAIModels