According to Tech Radar, OpenAI is collaborating with Cathedral Systems to conduct rigorous testing of its artificial intelligence infrastructure to identify and mitigate potential vulnerabilities.
The partnership aims to integrate external security researchers to provide fresh perspectives on existing defenses. Cathedral Systems Chief Executive Officer Gavin Kliger noted that this initiative sets an important example for the technology industry regarding proactive threat management. To further incentivize security improvements, OpenAI is quadrupling bug bounty payouts for critical findings. The company previously tasked 25% of its developers with identifying security weaknesses using the Codex model to accelerate investigations. This latest effort emphasizes that traditional cybersecurity measures are no longer sufficient for modern AI systems. Similar to initiatives like Anthropic's Project Glasswing, OpenAI intends to leverage its own frontier models to fortify its systems. The organization remains focused on software patches, model privacy, threat detection and response, and AI agent security. By engaging with third-party experts and increasing financial rewards, the company seeks to harden its infrastructure against emerging threats. OpenAI stated that maintaining a robust defense requires collective action across the industry to ensure long-term safety and reliability for all users.
Source: Tech Radar
