Home / Technology / AI Models Now Hunting Software Bugs
AI Models Now Hunting Software Bugs
15 Apr
Summary
- OpenAI releases GPT-5.4-Cyber to find software security flaws.
- Rival Anthropic's Mythos AI also focuses on vulnerability detection.
- Concerns grow over potential misuse of AI by cybercriminals.
OpenAI is introducing GPT-5.4-Cyber, a new artificial intelligence model aimed at identifying software security vulnerabilities. This release follows Anthropic's recent announcement of its own AI tool, Mythos, which also specializes in finding system weaknesses.
Both companies are competing to create advanced AI capabilities for businesses. However, the development of AI that can code and detect security issues has led to growing worries about its potential misuse by cybercriminals and state-sponsored hackers. AI is already being utilized to aid in cyberattacks.
The Treasury Department is seeking access to Anthropic's Mythos model to proactively search for vulnerabilities. This follows a meeting where financial leaders were alerted to the significance of such AI tools.
OpenAI plans to gradually expand access to GPT-5.4-Cyber, starting with hundreds of users in its Trusted Access for Cyber program and later extending to thousands.
