Security on the Path to AGI

OpenAI has updated its cybersecurity research support and bug bounty program to strengthen its security capabilities on the path to AGI.

The Cybersecurity Grant Program has expanded its scope of support, focusing on the following areas:

• Software patching: Vulnerability detection and patching using AI
• Model privacy: Preventing training data exposure and strengthening robustness
• Detection and response: Improving response capabilities against advanced persistent threats (APTs)
• Security integration: Enhancing accuracy and reliability of security tool and AI integration
• Agentic security: Strengthening defenses against sophisticated attacks on AI agents

In addition, OpenAI has introduced microgrants in the form of API credits so that researchers can quickly prototype innovative ideas.

The Security Bug Bounty Program has significantly expanded its reward scale. The maximum reward for discovering critical security vulnerabilities has been raised from the existing $20,000 to $100,000.

Meanwhile, OpenAI is building its cyber defense system using its own AI technology. Through AI-driven security agents, it is creating an intelligent security environment that can detect threats and respond quickly to evolving attack tactics.