OpenAI reveals how to re-align a misbehaving GPT-4o model

3 points| willmarquis | 8 months ago |openai.com