Anthropic releases Mythos preview, an AI model for defensive cybersecurity

Anthropic has unveiled a preview of Mythos, a new language model built specifically for cybersecurity tasks. It is the company's first public step toward narrowly specialized AI tools for protecting critical infrastructure, and a signal to the market that specialized models are gradually displacing general-purpose ones.

Mythos is being developed under a new Anthropic cybersecurity initiative. In the initial phase, access will go to a limited group of large companies with a high profile in information security. The model is meant to help specialists with defensive work: finding vulnerabilities, analyzing threats and building protective strategies before attacks can materialize. A preview launch is typical for Anthropic, which prefers iterative testing with trusted partners before a broad release.

The choice of defensive security is no accident. Anthropic consistently positions itself as a company that puts safety first. Mythos continues that logic: the model is tuned from the start not for offensive capabilities but for protecting systems. That distinction matters amid heated debate over the use of AI in military and intelligence operations. Anthropic's public position here is consistent: the company has signed voluntary commitments on safe AI development and takes an active part in dialogue with regulators.

Details on Mythos's technical capabilities are minimal for now. According to available information, the model underwent specialized training on data typical of the security field: code analysis for vulnerabilities, attack scenario modeling, threat classification and incident prioritization. Because the preview is closed, the full specification will come later, probably alongside case studies from the first partners.

The cybersecurity AI market is accelerating. Microsoft has integrated GPT-4 into Security Copilot, Google is developing Gemini-based tools for threat analysis, and CrowdStrike and SentinelOne are embedding LLMs in their SIEM platforms. By entering with a separate specialized model rather than simply integrating Claude into existing products, Anthropic signals its ambition to take a significant share of this segment. According to analysts, the AI market for cybersecurity will exceed $60 billion by 2028.

Launching through a partner model with a limited number of companies lets Anthropic collect quality feedback and refine Mythos on real-world tasks. This matters especially in cybersecurity: an AI error here can cost more than reputation. It can mean actual data compromise or a missed attack on critical infrastructure. The high stakes call for a cautious path to a public launch.

The arrival of Mythos is a sign of enterprise AI market maturity. Major players are increasingly moving from general-purpose models to versions specialized for particular industries: medicine, law, finance and now cybersecurity. For Anthropic this is a strategically important segment: corporate customers in critical industries are willing to pay a premium for reliability and specialization, which fits the company's enterprise-oriented business model well.