OpenAI's GPT-6 Astra Is So Powerful at Hacking, They're Worried
An AI that can hack at expert level is here — what does that mean for your data?
Deep Dive
GPT-6 Astra is OpenAI's most capable broadly deployed model and its first to reach the Critical level of cybersecurity capability under the Preparedness Framework.
Key Points
- GPT-6 Astra is OpenAI's first AI to reach 'Critical' cybersecurity ability, meaning it can hack at a near-expert level.
- It could be used to defend hospitals, banks, and apps by finding weaknesses before criminals do.
- OpenAI is withholding the most dangerous hacking features from the public and adding strict safety monitoring.
Why It Matters
An AI that can hack as well as experts could protect your personal data or become a powerful weapon — depending on who controls it.