Anthropic recently introduced its latest AI model, Mythos, which boasts advanced features tailored for defensive cybersecurity tasks. Concerns arose over Mythos’s extensive capabilities as the AI startup revealed uncovering numerous significant vulnerabilities across major operating systems and web browsers.
The launch of the model, named Claude Mythos Preview, was part of Anthropic’s controlled initiative, Project Glasswing, providing access to leading tech companies like Amazon, Microsoft, Nvidia, and Apple. Additionally, over 40 other organizations involved in critical software infrastructure were granted access.
Experts cautioned about Mythos’s potential to quickly identify and exploit unknown vulnerabilities, outpacing companies’ ability to address them. This could lead to an escalation in sophisticated cyberattacks, especially in sectors like banking that heavily rely on complex and interconnected legacy technology systems.
Anthropic highlighted the model’s capacity to detect software flaws at scale, acknowledging the severe risks it could pose to economies, public safety, and national security if misused. The unveiling of Mythos led to a decline in U.S. software stocks, underscoring concerns about AI advancements disrupting traditional businesses.
Regulators worldwide, including Bank of Canada Governor Tiff Macklem, emphasized the need to address the risks associated with rapid AI model advancements like Mythos. Discussions among financial sector stakeholders underscored the importance of preparing for the implications of such technologies on the global financial system.
While discussions with the White House and Anthropic CEO Dario Amodei focused on collaboration and cybersecurity, concerns were raised about balancing AI innovation with safety. Despite the Pentagon’s supply-chain risk designation on Anthropic, plans are in motion to offer a version of Mythos to major U.S. federal agencies.
Authorities in the U.S., Britain, and Germany are actively engaging with Anthropic and assessing the potential risks posed by Mythos. Financial institutions, particularly banks, are closely monitoring the situation and collaborating with regulators to address any vulnerabilities that could be exploited by the AI model.
Critics and industry insiders, including former White House AI and crypto czar David Sacks, echoed the need to take the concerns surrounding Mythos seriously. As coding models advance, the likelihood of discovering vulnerabilities increases, potentially enabling more sophisticated cyber exploitation, he emphasized.