OpenAI Launches GPT-6 Astra with Advanced Cybersecurity Capabilities
OpenAI released GPT-6 Astra on September 3, 2026, featuring autonomous cybersecurity functions.
Why it matters: Legal tech and corporate legal teams face new risks from GPT-6 Astra’s ability to find software vulnerabilities autonomously. This heightens the need for legal oversight, cybersecurity preparedness, and compliance with emerging AI regulations.
- GPT-6 Astra debuted on September 3, 2026, as OpenAI’s latest AI model with ‘Critical’ cybersecurity capabilities.
- Astra can autonomously detect and exploit unknown software vulnerabilities, increasing security risks.
- Access to Astra’s advanced cybersecurity features is highly restricted to select testers to prevent misuse.
- Initial availability is through OpenAI’s Daybreak Access program for enterprise users, with wider rollout planned.
- API pricing is $10 per million input tokens and $50 per million output tokens.
On September 3, 2026, OpenAI announced GPT-6 Astra, a significant advance in AI technology characterized by autonomous cybersecurity functions. This new model is the first OpenAI offering rated with 'Critical' cybersecurity capability, meaning it can independently discover and exploit software vulnerabilities unknown to traditional detection methods (VentureBeat).
OpenAI President Greg Brockman highlighted Astra’s ability to perform complex software engineering tasks autonomously, which will impact many sectors, including legal technology where automation of professional workflows is increasing (Fortune).
Because of Astra's capacity to autonomously exploit vulnerabilities, OpenAI has imposed stringent controls. Advanced cybersecurity functions are available only to a limited group of vetted testers, underscoring concerns about potential misuse or security breaches (Axios).
OpenAI CEO Sam Altman called for increased government oversight and international cooperation given the risks posed by rapidly advancing AI models like Astra, which could have broad regulatory and legal implications, especially related to cybersecurity, privacy, and compliance frameworks (Axios).
The release comes with access initially reserved for enterprise customers participating in OpenAI’s Daybreak Access program. Broader availability to ChatGPT Plus, Pro, Business, and Enterprise users is expected soon. Pricing for Astra’s API is set at $10 per million input tokens and $50 per million output tokens, positioning it as a premium offering for professional use (Investing.com).
This development marks not just a technological leap but also raises new legal and ethical challenges. Legal professionals will need to address evolving cybersecurity threats, potential liability around AI-driven vulnerabilities, and anticipate regulatory scrutiny as these tools enter widespread industry use.
By the numbers:
- September 3, 2026 — GPT-6 Astra launch date
- $10 per million input tokens — Astra API price
- $50 per million output tokens — Astra API price
Yes, but: While Astra’s cybersecurity capabilities are advanced, access is limited to reduce misuse risk, slowing widespread adoption of these features initially.
What's next: OpenAI plans to expand Astra access beyond initial enterprise testers to broader ChatGPT tiers soon, which may increase regulatory scrutiny and adoption in legal tech.