Anthropic preps Mythos 1 integration

Anthropic appears to be preparing a productized version of its Mythos cybersecurity model for Claude Code and Claude Security, according to leaked interface strings and code references reported on May 23 by TestingCatalog. The report cited a preview label — “claude-mythos-1-preview” — and a new string reading, “Access to the Claude Mythos model in Claude Code and Claude Security.” Anthropic has not publicly announced a Mythos 1 release. The company has said only that Mythos-class models could reach general release “in the near future” once “far stronger safeguards” are in place. ### Where did the Mythos 1 clues show up? TestingCatalog said users briefly saw “Mythos 1” in the Claude interface and that recent source-code changes pointed to integration with Claude Code and Claude Security. The same report described a redesigned Claude Security dashboard with vulnerability counts, seven-day and 30-day charts, and triage data for enterprise users. Anthropic’s own public statements stop short of confirming that rollout. (testingcatalog.com) In a Project Glasswing update cited by TestingCatalog, the company said it and its partners had found more than 10,000 high- or critical-severity vulnerabilities in essential software and added that Mythos-class models could be made more widely available after additional safeguards are built. ### What has Anthropic said publicly about Mythos so far? (testingcatalog.com) Anthropic said on April 7 that Claude Mythos Preview was being made available to selected partners through Project Glasswing, a program aimed at securing critical software and preparing the industry for more capable cyber models. In that technical post, the company said the model could identify and exploit zero-day vulnerabilities in major operating systems and web browsers and that more than 99% of the flaws it found had not yet been patched. (testingcatalog.com) Anthropic also said Mythos Preview had found bugs dating back decades, including a now-patched 27-year-old OpenBSD issue. Because most findings remain unpatched, the company said it was withholding technical details under coordinated disclosure practices. ### How does Claude Security fit into this? Anthropic said on April 30 that Claude Security entered public beta for Claude Enterprise customers. The product uses Claude Opus 4.7 to scan code for vulnerabilities and generate proposed fixes either on the Claude platform or through partner integrations. (red.anthropic.com) The April 30 announcement said Claude Security had already been tested by hundreds of organizations in a limited research preview and now includes scheduled and targeted scans, audit-system integrations, and tracking for triaged findings. (red.anthropic.com) Anthropic named CrowdStrike, Microsoft Security, Palo Alto Networks, SentinelOne, TrendAI and Wiz among technology partners embedding Opus 4.7 into their tools. ### Why are the 10,000-vulnerability claims getting attention? (claude.com) Anthropic said Project Glasswing and roughly 50 partners identified more than 10,000 high- or critical-severity vulnerabilities in system-critical software within a month. The Decoder, citing Anthropic’s update, reported that the pace of discovery now exceeds teams’ ability to verify, disclose and patch the flaws. (claude.com) The Decoder said Cloudflare reported 2,000 bugs, including 400 rated high or critical, while Mozilla said it found and fixed 271 vulnerabilities in Firefox 150 with Mythos Preview. The same report said Palo Alto Networks shipped five times as many patches as usual in its latest release and that Microsoft expected patch counts to keep rising. ### What happens next? Anthropic has not posted a launch date or pricing for Mythos 1. (the-decoder.com) The next concrete milestone remains any official update to Claude Code, Claude Security, or Project Glasswing from Anthropic, which has said wider Mythos-class availability depends on stronger safeguards and coordinated defensive preparation. (testingcatalog.com)