Anthropic has pulled back the curtain on a preview of its new frontier model, Mythos, deploying it with a select group of partners to tackle cybersecurity challenges. The AI startup, in a previously leaked memo, described the model as one of its “most powerful” to date.
Quick Facts
- Anthropic previews new frontier AI model, Mythos.
- Aimed at finding code vulnerabilities.
- Partners include Amazon, Apple, and Microsoft.
Project Glasswing: A New Alliance For Cybersecurity
The model’s limited debut is a core component of a new security initiative named Project Glasswing. This initiative brings together 12 major organizations, including Amazon, Apple, Broadcom, Cisco, CrowdStrike, the Linux Foundation, Microsoft, and Palo Alto Networks.
The collective goal is to use Mythos for “defensive security work,” primarily by scanning both proprietary and open-source software for critical vulnerabilities. These partners will share their findings to benefit the wider tech industry. While access is limited, Anthropic noted that 40 organizations outside the partnership will also get a look at the Mythos preview.
Mythos’ Early Wins and Capabilities
Mythos is a general-purpose model built for Anthropic’s Claude AI systems, boasting strong skills in agentic coding and reasoning. Although not specifically trained for security, its performance has been notable.
According to Anthropic, Mythos has already identified “thousands of zero-day vulnerabilities, many of them critical” in just a few weeks of testing. The company added that many of these security flaws are one to two decades old, lurking undetected in widely used software.
From ‘Capybara’ Leak to Official Preview
News of the powerful model first surfaced last month through a data security incident. A draft blog post, found in an unsecured cache of documents, referred to the model by its codename, “Capybara.”
The leaked document stated, “‘Capybara’ is a new name for a new tier of model: larger and more intelligent than our Opus models.” It described the model as “by far the most powerful AI model we’ve ever developed,” with performance in software coding and cybersecurity far exceeding its publicly available models. The leak also highlighted the potential threat if the model were weaponized by malicious actors to exploit bugs rather than fix them.
What This Means for MENA’s Tech Ecosystem
For the rapidly digitizing MENA region, the introduction of an AI model like Mythos is significant. As governments and private enterprises across the GCC and beyond invest heavily in digital transformation, their attack surface for cyber threats expands. Startups building on cloud infrastructure and enterprises managing legacy systems are both prime targets.
An AI tool that can proactively identify decades-old vulnerabilities in foundational software could be a game-changer for regional cybersecurity firms and corporate security teams. It offers a way to get ahead of threats and secure the digital infrastructure underpinning initiatives like Saudi Vision 2030 and the UAE’s digital economy strategy. The collaboration between Anthropic and tech giants like Microsoft and Amazon, both of whom have a major cloud presence in the region, suggests that the benefits of this technology could eventually trickle down to their MENA-based customers.
About Anthropic
Anthropic is an AI safety and research company that’s working to build reliable, interpretable, and steerable AI systems. The company is focused on developing advanced AI models and conducting research to ensure that AI technology is developed responsibly and benefits humanity.
Source: TechCrunch
