EU enters talks with Anthropic over AI model Mythos risks
The EU is now asking questions. And the answers could shape the future of AI safety worldwide.
European Union – (Web Desk) – The European Union is in talks with Anthropic. The US AI company has built a model that worries even its own creators.
The model is called Claude Mythos. It is very good at finding weaknesses in software. That makes it a potential tool for hackers. Because of this, Anthropic has delayed its full public release.
The EU’s European Commission confirmed the discussions on Thursday. Spokesman Thomas Regnier told reporters: “We have a new AI model being released. It comes with risks. We need information about those risks.”
Regnier added that a first meeting with Anthropic took place on Wednesday. More meetings are planned.
Anthropic says Mythos can scan huge amounts of code on its own. It finds hidden security flaws across all types of software. This includes operating systems and web browsers.
The real danger? It works at a speed no human can match. Experts warn it could take down banks, hospitals, or national infrastructure within hours.
Japan commits $2 million in aid for Pakistan’s climate-affected communities
Anthropic limited Mythos to just 40 major tech companies. The goal is to let them fix vulnerabilities before attackers can exploit them. But no foreign companies were included. That has raised concerns about global preparedness.
Some experts are skeptical. They point out that Anthropic — a player in a fierce AI race — could benefit from the hype around a “dangerous” model.
Still, the concerns are being taken seriously at the highest levels. The heads of America’s biggest banks recently met with Federal Reserve Chairman Jerome Powell and Treasury Secretary Scott Bessent. They discussed the security risks of Mythos before it is even fully released.
Comments are closed, but trackbacks and pingbacks are open.