OpenAI Says It Expects That Future Models Will Be Able To Identify & Exploit Vulnerabilities Of Which It Is Not Yet Aware

Benzinga · 2d ago

https://openai.com/index/securing-our-frontier-model-training-environment/