As the "Mythos" large model, which has drawn significant attention from the cybersecurity industry, is about to be publicly released, AI giant Anthropic announced a major breakthrough on Tuesday. The company has decided to grant preview access to this cutting-edge model to approximately 150 new institutions around the world, aiming to comprehensively scan and fix security vulnerabilities in global critical cyber infrastructure in advance.

As a large model with disruptive capabilities in cybersecurity detection, Mythos has already demonstrated remarkable destructive power and constructive potential during early testing. In the first round of previews opened in April to 50 initial partners, the model successfully identified over 10,000 high-risk or severe system security flaws for these organizations.

Accelerating the Repair of Critical Infrastructure

More excitingly, the Mythos large model detects vulnerabilities ten times faster than traditional manual methods or tools. This technological breakthrough allows cybersecurity professionals to reinforce their defense systems at an efficiency previously unimaginable, before hackers launch attacks.

The 150 new institutions added to the preview list mostly control the core interests of their respective countries, covering key infrastructure industries such as power, water supply, healthcare, communications, and hardware. Since a successful attack on these institutions' codebases could lead to catastrophic consequences, using AI for proactive defense has become particularly urgent.

Public Release Is Now on the Agenda

In addition to having top-tier vulnerability discovery capabilities, the Mythos large model also possesses industry-leading programming and code analysis skills. However, as the model can easily identify a large number of vulnerabilities, the current bottleneck in cybersecurity has shifted to how quickly these emerging defects can be verified and disclosed.

According to a previous announcement from Anthropic when releasing Claude Opus 4.8, the Mythos large model is expected to be officially available to all public users in the coming weeks. With the commercialization of this top-tier model, the future landscape of cybersecurity and the AI agent ecosystem will undergo profound changes.