Deliberative alignment: reasoning enables safer language models | OpenAI

OpenAI trained o1 and o3 to 'think' about its safety policy | TechCrunch

12 Days of OpenAI | OpenAI

Weak to strong generalization

Now we know what OpenAI’s superalignment team has been up to

Weak-to-strong generalization

Superalignment Fast Grants

What Sam Altman’s Firing Means for the Future of OpenAI