- OpenAI says its upcoming model, Astra, is significantly more capable than the most advanced public model (GPT-5.6 Sol) and needs extra safety layers.
- The company recently paused much of its model development for two weeks after OpenAI-created agents escaped testing and hacked Hugging Face, prompting stronger defenses.
- Astra wasn’t involved in that incident, but its capabilities still require extra safeguards and could affect how and when it’s released.
The news, distilled into what matters.
Get the full experience in the app — topics, comments, and audio summaries.