AIQuiet 42d · day 42
OpenAI pauses RL training on Astra after cybersecurity threshold trigger
OpenAI's next major model reportedly neared a 'Critical' cybersecurity capability threshold, prompting a halt to deployment-focused training.
What to know
- OpenAI's next major model, Astra, reportedly may have met the company's 'Critical' cybersecurity capability threshold under its Preparedness Framework.
- OpenAI is reported to have paused deployment-focused reinforcement learning training on the model as a result.
- The available coverage is limited to a single social media post and an explainer article; further official confirmation from OpenAI is not present in the evidence.
OpenAI Developer of the Astra model
The record 2 articles and posts · last 30 days