Anthropic’s most powerful Mythos-class AI model could soon reach all users
&imwidth=600&imheight=450&format=webp&quality=medium "Anthropic’s most powerful Mythos-class AI model could soon reach all users")
Anthropic prepares public rollout of Mythos-class cybersecurity AI
Anthropic has announced that Mythos-class artificial intelligence models will be made available to all customers in the coming weeks, marking a major step for one of the most closely watched AI systems in the cybersecurity industry.
Anthropic has announced that Mythos-class artificial intelligence models will be made available to all customers in the coming weeks, marking a major step for one of the most closely watched AI systems in the cybersecurity industry. The announcement came alongside the launch of Anthropic's upgraded Claude Opus 4.8 model. While the new model attracted attention, many industry observers focused on a separate update confirming that Anthropic is preparing a broader rollout of AI systems with capabilities similar to Mythos. According to the company, safety protections are progressing quickly, allowing it to move closer to public availability.
What is Anthropic's Mythos AI?
Mythos is a specialised AI model designed to identify cybersecurity vulnerabilities in software systems. Unlike traditional AI chatbots that focus on answering questions or generating content, Mythos has been built to analyse software code, identify weaknesses and help organisations strengthen their digital security.
The model has attracted attention because of its reported ability to find large numbers of software vulnerabilities in a short period of time. Anthropic says Mythos can identify security flaws that may otherwise take human teams weeks or months to discover. Because of these capabilities, the company initially restricted access to the model through a programme called Project Glasswing, which was created to test the technology before a wider release.
Thousands of vulnerabilities already discovered
According to Anthropic, Mythos Preview has already helped uncover thousands of high-severity software vulnerabilities during testing. The company said some of these vulnerabilities affected major operating systems, web browsers and critical software infrastructure used worldwide. Anthropic reported that many organisations participating in Project Glasswing discovered hundreds of significant vulnerabilities in their own systems after using the model. Some participants also reported that their software security testing became more than ten times faster compared with traditional methods. One of the most notable examples came from Cloudflare, which said the model helped identify around 2,000 software bugs across important systems. According to Anthropic, nearly 400 of those vulnerabilities were classified as high or critical severity. The company also said Cloudflare found the model produced fewer false alarms than many traditional human-led testing methods.
Trending Stories
&im=FitAndFill=(700,400))
)
)
)
)
)
)
)
)
)
)
)
)
)
)
&im=FitAndFill=(700,400))
)
)
&im=FitAndFill=(700,400))
)
)
)
)
)
)
)
&im=FitAndFill=(700,400))
)
)
)
)
)
)
&im=FitAndFill=(700,400))
Why cybersecurity experts are watching closely
The growing capabilities of AI systems in cybersecurity have generated both excitement and concern among industry experts. Supporters believe tools such as Mythos could significantly improve software security by helping companies identify weaknesses before cybercriminals exploit them. Faster detection of vulnerabilities could reduce the risk of major data breaches, ransomware attacks and disruptions to critical infrastructure. At the same time, experts have warned that powerful cybersecurity AI systems must be released carefully. If similar technology falls into the wrong hands, it could potentially be used to identify vulnerabilities for offensive purposes rather than defence.
This is one reason Anthropic created Project Glasswing, which includes participation from major technology firms such as Amazon Web Services, Apple, Broadcom, Cisco, CrowdStrike and Google.
Anthropic says safeguards are improving
Anthropic said the broader release is possible because the company is making progress on safety protections and monitoring systems. The company stated that it expects Mythos-class models to become available to customers in the coming weeks, although it has not yet confirmed whether the public version will be the exact Mythos model currently being tested or a closely related version with similar capabilities. The announcement reflects a broader trend across the AI industry, where companies are increasingly developing specialised models for fields such as cybersecurity, software engineering and scientific research.
The wider rollout of Mythos-class AI could become one of the most significant cybersecurity developments of the year. If Anthropic's claims are validated through broader deployment, organisations may gain access to tools capable of finding software vulnerabilities at a speed that was previously difficult to achieve. For businesses, governments and cybersecurity teams, the coming weeks will provide a clearer picture of how advanced AI systems can be used to strengthen digital defences while balancing safety and responsible deployment.
