Home / Business and Economy / US Banks, Treasury Tackle AI's Cyber Threat
US Banks, Treasury Tackle AI's Cyber Threat
10 Apr
Summary
- Treasury secretary met with top US bank leaders.
- Discussion focused on AI's advanced cyber vulnerability detection.
- Anthropic's new AI model can find thousands of severe vulnerabilities.
US Treasury Secretary Scott Bessent recently met with executives from leading US banks to address the cyber risks associated with Anthropic's latest AI model. The discussion involved heads of Bank of America, Citigroup, Goldman Sachs, Morgan Stanley, and Wells Fargo. Federal Reserve Chair Jay Powell also attended the meeting, highlighting the significance of the concerns.
The core issue revolves around Anthropic's new model, dubbed Claude Mythos Preview, which possesses advanced capabilities in identifying cybersecurity vulnerabilities. This has triggered apprehension among government officials and financial leaders about its potential misuse by malicious actors.
Anthropic announced the model's release to select partners like Amazon, Apple, and Microsoft, aiming to provide them a 'head start on being able to secure vulnerabilities.' The company stated that AI models have reached a coding proficiency surpassing most humans in finding software flaws, with Mythos already uncovering thousands of severe vulnerabilities across major operating systems and web browsers. Anthropic has also engaged in discussions with US officials regarding the model's offensive and defensive cyber capabilities. This development follows previous data leaks from Anthropic, which the company attributed to human error, further fueling security discussions.
