Keep pulling the thread on Scott Besant.
US Treasury Secretary Scott Besant convened Wall Street bank leaders to warn them about Anthropic's new AI model, Mythos.
US Treasury Secretary Scott Besant warned that Anthropic's Mythos AI model could enable a new era of cataclysmic cyber attacks carried out autonomously by bots.
Anthropic's Mythos model possesses the ability to autonomously discover software vulnerabilities on a scale not previously seen.
During testing, Anthropic's Mythos model found 181 ways to exploit the Firefox browser, compared to only two exploits found by its predecessor, Cloud Opus 4.6.
The capability of Anthropic's Mythos model to chain together minor exploits into a significant, autonomous attack campaign is the primary concern for security researchers.
The US Treasury Secretary encouraged banks to use Anthropic's Mythos model to identify vulnerabilities in their own systems.
A week after the Mythos news, OpenAI released a model named GPT-5-4 Cyber to demonstrate it has a model with comparable cybersecurity capabilities.
Michael Dung predicts that as AI capabilities continue to advance without plateauing, discussions about government intervention, including potential nationalization of AI firms, will intensify.
European policymakers are concerned that a critical AI capability developed in a less collaborative jurisdiction like China would not be shared with Western banks to help them prepare defenses.
Anthropic has stated its Mythos model is so powerful that the company is only making it available to a select group of firms.
Anthropic's previous model, Cloud Opus 4.6, demonstrated the ability to find 500 zero-day exploits.
Michael Dung argues for stronger, uniform cybersecurity regulations for banks, as self-regulation may fail due to varying risk appetites, similar to what occurred before the 2008 global financial crisis.