AI cybersecurity is increasingly competitive, with OpenAI and Anthropic leading the way. OpenAI is launching an advanced cybersecurity product for a selected group of partners, while Anthropic is implementing Project Glasswing, which focuses on identifying critical software vulnerabilities before malicious actors do. Both initiatives signify a shift toward specialized AI tools with both offensive and defensive capabilities, raising concerns about accountability in case of failures.
Anthropic’s Project Glasswing shows the potential of AI tools, having identified thousands of vulnerabilities, including long-standing flaws in major software. A staggering 72% increase in AI-driven cyberattacks is reported annually, with 87% of organizations anticipating AI-related incidents by 2025. The dual-use nature of these AI tools can mean they find vulnerabilities for both protection and exploitation, complicating regulatory responses. Thus, the controlled rollout strategies of OpenAI and Anthropic are crucial in addressing the risks associated with AI in cybersecurity.
Source link
