What is Claude Mythos?

Claude Mythos is one of Anthropic’s latest AI models, developed as part of its broader AI system called Claude — it includes the company’s AI assistant and a family of models that rival OpenAI’s ChatGPT and Google’s Gemini.

Anthropic revealed the model in early April as a ‘Mythos Preview’ and described it as strikingly capable in computer security tasks, but Red-team researchers, who test AI models on specific requests, found that Mythos could locate dormant bugs in decades-old code and exploit them with ease.

Due to these capabilities. Anthropic did not make Mythos widely available to its users; Instead, it granted access to 12 tech companies through an initiative called Project Glasswing — this project aims to strengthen resilience against Mythos and secure the world’s most critical software.

The companies involved in Project Glasswing include major tech giants such as Amazon Web Services, Apple, Microsoft, Google, Nvidia, Broadcom, and Crowdstrike, but Anthropic also provided access to over 40 organizations responsible for critical software.

Why Are There Concerns?

Anthropic claims that during testing. Mythos outperformed humans in cyber-security and hacking tasks; it can identify critical bugs in old systems without much oversight and suggest ways to exploit them. One such vulnerability had been present in a system for 27 years.

Certainly, it is serious enough to warrant the attention of all the finance ministers, said Canadian finance minister François-Philippe Champagne, referring to discussions about Mythos at an International Monetary Fund meeting in Washington DC.

Bank of England boss Andrew Bailey told the BBC that the institution is carefully examining the potential risks that the latest AI development could pose to cyber crime. Meanwhile, the EU has also been in discussions with Anthropic about its concerns regarding Mythos.

Anthropic’s CEO, Dario Amodei, said the company is willing to work with US government officials to help defend against the risks associated with these models. In a video released with Project Glasswing’s launch, Amodei emphasized the need for collaboration to address potential threats.

What Have Cyber Experts Said About It?

Ciaran Martin, former head of the UK’s National Cyber Security Centre, stated that the claim that Mythos could uncover critical vulnerabilities more quickly than other AI models had ‘really shaken people.’

He added that Mythos is essentially a ‘really good hacker’ and can exploit existing weaknesses that organizations might not have patched against. However, many independent cyber-security analysts have not yet tested Mythos themselves and remain skeptical about its performance.

The UK’s AI Safety Institute recently concluded that while Mythos is a very powerful model, its biggest threat would be against poorly defended, vulnerable systems. In theory, if basic cyber-security measures are in place, this model would be stopped.

Fears relating to AI are not new, and the development of new models and tools is frequent. These often come with promises of revolutionizing our lives, for better or worse. Capitalizing on this mix of fear and excitement has become a hallmark of the AI sector and its marketing strategies.

In the case of Mythos, we still do not know enough to determine whether these hopes or fears are justified or more a reflection of the hype surrounding the industry. According to the National Security Cyber Security Centre, the most important thing we can do now is not panic and instead focus on ensuring basic cyber-security is in place.

Most hackers do not need super AI tools to breach systems when much simpler attacks often suffice. Martin told the BBC that while some see this as an apocalyptic event, others believe it is largely hype. He added that whether it was this tool or subsequent ones made by Anthropic or its rivals, there was an opportunity to build a safer online world.

In the medium-term, there’s an opportunity to use these tools to fix a lot of the underlying vulnerabilities in the internet, Martin said.