The introduction of Claude Mythos, an advanced AI model from Anthropic, has raised significant concerns about its potential risks to digital security. The technology firm claims that Mythos can surpass human capabilities in certain hacking and cybersecurity activities. This revelation has sparked discussions among regulators, financial institutions, and legislators regarding the implications for online services.
Mythos, part of Anthropic's Claude system, was first unveiled in early April as "Mythos Preview." Its abilities have been validated by researchers who noted its effectiveness in identifying and exploiting vulnerabilities within legacy code. To mitigate risks associated with the model, Anthropic has restricted access to it, providing it only to 12 selected companies through an initiative known as Project Glasswing, which aims to bolster the security of critical software.
Participants in this initiative include major players such as Amazon Web Services, Apple, Microsoft, and Google. Additionally, over 40 organizations responsible for essential software have gained access. Anthropic's CEO, Dario Amodei, has expressed willingness to collaborate with US government officials to address the challenges posed by AI advancements like Mythos.