OpenAI recently introduced a new cybersecurity model called ChatGPT Cyber, designed to enhance the capabilities of cybersecurity professionals within its Daybreak program. This release follows the discovery of a critical vulnerability in the widely used Ubuntu Linux operating system by Max Justicz, CTO of Charge Robotics, who used an earlier OpenAI model, GPT 5.6 Sol, to identify the flaw. His findings highlight how powerful AI tools can uncover complex bugs much faster than humans, raising concerns about the widening gap between hackers and defenders as AI becomes more accessible.

The Daybreak program now offers two tiers: Daybreak Blue for defenders and Daybreak Red for offensive security researchers. OpenAI has used the new model to scan major open-source projects, identifying nearly a thousand potential issues in the Linux kernel and discovering dozens of previously unknown Firefox vulnerabilities. The company plans to expand Daybreak access to a broader range of participants, aiming to democratize advanced security research and assist critical infrastructure maintenance.

Despite these efforts, cybersecurity experts caution the challenge remains immense. Justicz pointed out that the AI tools available to defenders have recently been restricted, while malicious actors have access to less regulated, more powerful AI models. He also highlighted the significant costs involved in effectively using these AI tools, noting his own expenses of around $5,000 to find exploits. The relatively small number of vetted defenders contrasts sharply with the many potential attackers who could misuse open AI models, fueling concerns about the race between securing and attacking digital infrastructure.

OpenAI has committed $1 billion to support cybersecurity professionals working on critical infrastructure, emphasizing the urgent need to patch vulnerabilities before AI-driven exploits become widespread. The company is dedicating significant resources toward vulnerability discovery and remediation, offering pro bono services such as the recent Linux kernel audit. However, compared to other high-cost AI projects, the investment in cybersecurity is modest, suggesting that more urgency and resources may be required to safeguard the foundational technologies that underpin the digital world.