Tag
A blog post analyzing Anthropic's recent LLM-assisted cryptanalytic attacks on HAWK and reduced-round AES, arguing that LLMs will not break established symmetric cryptographic schemes.
The tweet reports that serious cyber vulnerability disclosures are climbing sharply, with 21 major tech organizations publishing about 2,500 high- and critical-severity CVEs in July — roughly 5× the previous monthly record — following Anthropic's reveal that Claude Mythos Preview could autonomously find software vulnerabilities.
Anthropic's unreleased Claude Mythos Preview model discovered novel attacks on the HAWK post-quantum signature scheme and a reduced-round AES variant, significantly reducing attack costs and potentially impacting standardization.
A spike in high- and critical-severity CVE disclosures followed Anthropic's release of Claude Mythos Preview, which can autonomously discover software vulnerabilities, leading to a 3.5x increase in monthly records.
This article explores the reasons behind the U.S. government's initial restriction on the AI model Claude Mythos and its subsequent decision to allow it, focusing on regulatory shifts and AI safety considerations.
Zhipu AI has released a new model that reportedly matches the performance of Claude Mythos in identifying security vulnerabilities.
22-year-old developer Kye Gomez reversed Anthropic's Claude Mythos black box architecture in just two days and open-sourced the OpenMythos project, using Recurrent-Depth Transformer and other techniques, achieving performance equivalent to a 1.3B model with 770M parameters.
Trained a 9B model (Qwythos 9B) on Claude Mythos traces, achieving strong results in bug-finding, SQL, and function calling while running locally on an A6000 with llama.cpp.
An article arguing that Anthropic CEO Dario Amodei's policy statements advocating for government power to block AI deployment directly led to the US government restricting access to Anthropic's latest models, Claude Fable and Claude Mythos.
A developer shared a free 12-minute tutorial on using Claude Mythos to build award-winning animated websites.
Anthropic announces a 30-day data retention policy for Mythos-class models (Claude Mythos 5 and Fable 5) for safety and misuse detection, affecting enterprise and zero-data-retention customers starting June 9, 2026.
Anthropic expands Project Glasswing, utilizing its Claude Mythos AI model, to critical infrastructure organizations across 15+ countries to identify zero-day vulnerabilities.
Anthropic's Claude Mythos and OpenAI's ChatGPT-5.5 frontier models raise cybersecurity concerns due to their ability to autonomously identify and exploit vulnerabilities. Researchers from the Max Planck Institute discuss the real risks and the need for pooled European knowledge on offensive AI systems.
Japanese banks are getting early access to a new OpenAI model for security testing, reportedly comparable to Anthropic's Claude Mythos.
Anthropic's new AI model Claude Mythos, using the Claude Code framework, reportedly solved Erdős's distinct distances problem by finding alternative simple proofs, following OpenAI's earlier disproof. This demonstrates LLMs' ability to make independent scientific breakthroughs.
Anthropic's new AI model, Claude Mythos, identified over 10,000 high and critical security flaws in global system software within a month, with a false positive rate better than human testers, significantly advancing AI-driven cybersecurity.
Anthropic's Project Glasswing has used Claude Mythos Preview to find over 10,000 high or critical severity vulnerabilities in critical software, with partners like Cloudflare reporting a tenfold increase in bug finding rates, highlighting the shift from discovery to patching as the bottleneck.
Claude Mythos, a new Anthropic AI model, has lost its preview label on Google Cloud's Vertex AI, signaling imminent public release; this transition indicates enterprise-grade approval and rapid distribution to Fortune 500 companies.
Three security researchers bypassed Apple's MIE protection in six days using AI model Claude Mythos, demonstrating a rare exploit that requires no traditional malware and gains root access on macOS. The full paper will be published after Apple releases a patch.
The thread discusses recent evidence that AI agents have become largely autonomous, with Claude Mythos solving previously unsolved cyber attack simulations and exceeding current benchmark measurement limits, indicating super-exponential progress. It highlights the security implications and institutional responses.