Technology designed for cybersecurity can be a cyber threat, admits its creator. Global financial authorities are worried.
Anthropic launched this month the Mythos and presented it as its most advanced artificial intelligence model yet, but that’s not necessarily good news.
Designed for defensive cybersecurity tasks, the “newborn” Mythos has already begun to fuel a wave of concern, namely by potentially surpassing the response capacity of traditional security teams.
According to Anthropic, the model, in preview, managed to identify thousands of serious vulnerabilities in practically all major operating systems and web browsers. The company itself warns that this capability, although it can be used to reinforce digital defense, could also represent big risks to the economy, public safety and national security if it is abused.
Access to the Claude Mythos Preview was initially limited through a controlled initiative called “Project Glasswing,” according to . Organizations with access to AI include technology giants such as Amazon, Microsoft, Nvidia and Apple, as well as more than 40 entities responsible for developing or maintaining critical software infrastructures.
Fears mainly focus on the possibility that Mythos will be able to discover and exploit previously unknown flaws at a faster rate than companies can fix them.
Experts emphasize that advanced programming capabilities and technology autonomy can accelerate sophisticated cyberattacksespecially in sectors such as banking.
The impact of these concerns has already been felt in the markets. On April 9, shares of North American software companies fell following the launch of Mythos on April 7, reigniting fears about the disruptive effect of AI on traditional technology companies.
In the United States, the White House met with Anthropic CEO Dario Amodei to discuss cooperation, cybersecurity and the balance between innovation and security — this happened despite the Pentagon giving Anthropic a formal supply chain risk designation. According to Bloomberg, the US government is also preparing to make a version of Mythos available to large federal agencies.
Europe is also starting to worry. In the United Kingdom, authorities opened talks with banks and cybersecurity officials. Australia and South Korea are also following the new artificial intelligence model with increasing concern.
For now, no formal regulatory measures have been announced.
