SecurityNov 3, 2026, 11:17 PM UTC·2 min read

OpenAI Raises Bug Bounty Rewards to $172,000 in Security Push

OpenAI is raising its top bug bounty payout to $172,000 and teaming up with security startup Cathedral Systems to test its own systems against advanced attacks.

Illustration of a digital shield and padlock over code, symbolizing OpenAI's expanded security testing and bug bounty program. — Image: اليوم السابع

SUMO

Dialo

ediviewer

OpenAI has increased the maximum reward in its Bug Bounty program to $172,000, including additional incentives, for exceptional discoveries. According to the update, that is more than four times the previous payout for critical vulnerabilities.

The move comes alongside a new partnership with cybersecurity startup Cathedral Systems. OpenAI says the collaboration is part of an intensive internal security testing effort that uses AI to simulate attacks and uncover weaknesses that conventional scans may miss, so they can be fixed before attackers exploit them.

Who is Cathedral Systems

Cathedral Systems was founded by former members of the DOGE department. It drew attention after its AI agents reportedly found security flaws in external business systems belonging to Anthropic in under 12 hours, according to circulating reports. OpenAI aims to draw on that capability to test its own digital defenses.

Lessons from earlier testing

The company has done this before. In one earlier round of intensive security testing, OpenAI assigned about 25% of its product engineers to hunt for and fix weaknesses. The engineers used Codex to analyze security reports, re-test discovered vulnerabilities in secure environments and develop the software updates needed to address them.

OpenAI stresses that security testing is not a temporary campaign but an ongoing process of reviewing systems and updating protections as new threats emerge.

Why it matters

As AI tools get better at finding and exploiting vulnerabilities and at analyzing code, companies are under pressure to use the same technology defensively. OpenAI is betting that pairing human researchers with AI models will surface problems faster and shorten the gap between a flaw appearing and being fixed.

For independent researchers, the higher rewards are an incentive to report issues under the program's rules. For the wider industry, the step shows that protecting internal systems and data is now central to building trustworthy AI services, especially as the risk of AI-assisted cyberattacks grows.