Anthropic unveils Claude Mythos model, finds thousands of OS vulnerabilities via Project Glasswing
Anthropic has unveiled Claude Mythos, a new AI model designed for cybersecurity that has already discovered thousands of high-severity vulnerabilities in every major operating system and web browser. The model is being distributed as a preview to over 40 organizations and major technology partners including Apple, Google, Microsoft, and Amazon Web Services through Project Glasswing, a coordinated cybersecurity initiative.
Anthropic Unveils Claude Mythos Model, Discovers Thousands of Critical Vulnerabilities
Anthropric has announced Claude Mythos, a specialized AI model designed for cybersecurity vulnerability discovery, alongside Project Glasswing—a multi-organization initiative to secure critical software infrastructure. The model has already identified thousands of high-severity vulnerabilities across every major operating system and web browser, some of which survived decades of human review and millions of automated security tests.
Project Glasswing Partnership
Project Glasswing brings together 12 founding technology companies and organizations: Amazon Web Services, Anthropic, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorganChase, the Linux Foundation, Microsoft, NVIDIA, and Palo Alto Networks. An additional 40+ organizations that develop or maintain critical software have access to the Claude Mythos Preview model.
The initiative's stated goal is to identify and remediate security vulnerabilities before broader deployment, acknowledging that AI-powered vulnerability discovery capabilities will proliferate among actors—not all of whom may deploy them safely.
Capabilities and Performance
Claude Mythos demonstrates improvements over Anthropic's Claude Opus 4.6 in reasoning, agentic search, computer use, and notably agentic coding. One example Anthropic provided involved discovering and chaining together a Linux kernel vulnerability that could enable complete machine compromise.
Anthropic's system card indicates the model shows measurable gains in vulnerability detection compared to prior generations, though specific benchmark scores were not disclosed in the announcement. The model's ability to identify flaws that persisted through extensive human and automated review suggests substantially improved detection capabilities.
Availability and Future Plans
Anthropic explicitly stated it does not plan to make Claude Mythos Preview generally available to the public. However, the company indicated its "eventual goal is to enable our users to safely deploy Mythos-class models at scale—for cybersecurity purposes, but also for the myriad other benefits that such highly capable models will bring."
This suggests a phased approach: controlled preview access to vetted organizations now, with broader availability contingent on developing safety protocols for deployment at scale.
What This Means
Anthropic is deploying its most capable model not for general use but specifically for defensive security purposes—positioning itself as a leader in AI safety through controlled capability release. The coordinated Project Glasswing approach represents an acknowledgment that frontier AI capabilities for vulnerability discovery are inevitable, and that proactive coordination among major technology companies is necessary to ensure such tools are deployed defensively before they proliferate to less responsible actors.
For organizations in the preview program, access to Mythos provides immediate security benefits. For the broader industry, Project Glasswing sets a precedent for how companies might manage the release of potentially dual-use AI capabilities.
Related Articles
Anthropic mathematician Levent Alpöge claims Claude-assisted proof of the Riemann hypothesis, posts it in Lean
Levent Alpöge, a mathematician employed by Anthropic, says he has proven the Riemann hypothesis, first proposed in 1859, working with Claude. He posted the proof coded in Lean, and experts have yet to confirm that it is correct.
Anthropic Adds Experimental Code Mode to Claude Code v2.1.323, Chaining Tool Calls in JavaScript
Anthropic has added an experimental Code Mode to Claude Code v2.1.323. Claude can write JavaScript that chains Bash, Read and Edit calls in one step, and keep talking while long commands run.
Thinking Machines Lab releases Inkling 2 Mini: 26B-parameter looped MoE with 3B active and 1M-token context
Thinking Machines Lab has released Inkling 2 Mini, a multimodal Mixture-of-Experts model with about 26B physical total parameters and 3B active per token. It repeats its middle layers three times, supports a 1M-token context, and ships with a DSpark drafter. Hugging Face reports it leads 32 of 48 benchmarks against four comparable models.
NVIDIA releases Nemotron-Labs-Sigma: 14 open 3B and 8B continuous diffusion language model checkpoints
NVIDIA released Nemotron-Labs-Sigma, a set of 14 open continuous diffusion language model checkpoints at 3B and 8B parameters. According to MarkTechPost, Sigma denoises text in a learned 16-D embedding space, distills to few-step samplers, and decodes up to 4.20x faster than autoregressive decoding.
Comments
Loading...