AI Generated Image
OpenAI Is About to Release Its First AI Model With ‘Critical’ Cyber Abilities
WIRED • • business
Key Points:
- OpenAI announced its upcoming AI model, Astra, has reached the company’s "critical" cybersecurity capability threshold, meaning it can independently find and exploit previously unknown software vulnerabilities.
- OpenAI plans to release Astra publicly soon but will restrict its advanced cyber capabilities to select partners through its Daybreak Blue early-access program to ensure safe deployment.
- The company paused Astra’s development temporarily to implement additional safety and security controls, resuming training only after confirming it can release the model safely.
- Astra can chain multiple exploits to penetrate deeper into target systems, raising concerns about AI-driven cyber risks; OpenAI is using a “misalignment monitor” to prevent misuse and limit user access to dangerous capabilities.
- OpenAI is collaborating with digital infrastructure partners and government agencies to responsibly manage Astra’s capabilities and help strengthen cybersecurity defenses before broader release.
