OpenAI has built an LLM super-hacker called GPT-Red that it uses as a sparring partner to help its other models boost their ...
Anthropic is hiring AI safety experts to stop its models from being misused for nuclear threats, explosives, cybercrime, ...