Withholding a model has been a long-standing lever in Frontier Lab safety plans. Before Mythos Preview and Glasswing, OpenAI launched their Trusted Access for Cyber program for GPT-5.3-Codex (their first model to reach “High” cybersecurity capability). Anthropic have now launched their similar Cyber Verification Program.
Mythos enters the chatThe UK AI Security Institute (AISI) put it best with this summary, “Mythos Preview represents a step up over previous frontier models in a landscape where cyber performance was already rapidly improving”. The AISI recently created an evaluation which tests model capability on a network attack simulation spanning 32 stages of an attack chain (estimated to take a human 20 hours to complete). Mythos Preview is the first model to solve this challenge from start to finish, succeeding on 3 of 10 attempts with a 100 million token budget. AISI expect greater budget would improve results further. Mytho
Mythos enters the chatMozilla says it has patched over 400 bugs with the help of Anthropic Mythos The tool had to be 'harnessed' to Firefox for the best results Mythos offers a step away from the typical 'unwanted slop' provided by AI fuzzing tools Anthropic’s latest cybersecurity tool Mythos is continuing to make waves… …
… That means extending identity controls to non-human identities, enforcing least-privilege access and setting clear boundaries around what AI systems can reach and do. Data security - For many organizations, the core risk is not only what AI tools can produce but what it can access, expose or move. …
… What I found was less a deliberate strategy than an honest reflection of how fast things had moved: AI tools embedded in workflows, some vendor-approved, some not, adopted by smart people solving real problems faster than policy could keep up with. Chief Information and Security Officer, Smartsheet. …
… Security The UK wants to use AI to defend against Russia and China's "relentless attacks" on critical infrastructure. Security Numerous legitimate tools being used in RMM attacks against Brazilian targets. Security If a remote session fails, hackers will come to install malware in person. …
… AI as the differentiator for defenders Some security experts have suggested that true cybersecurity tradecraft is found in chaining everything together, or evading discovery, or they emphasize that the human was still needed. …
… Simple and complex threats MSHTA is an old tool that was originally designed for lightweight desktop and administrative tasks but is, as many other legacy tools, being abused to run malicious scripts, download malware, or bypass security controls. “Since the start of the year, we have observed an i… …
… Threat actors are increasingly using AI tools to craft highly convincing phishing emails and scams that can convince even the most wary of internet users. …
… Pro Hackers hide ransomware tools inside virtual machines using QEMU, allowing attacks to bypass security controls and remain largely invisible. …
… Pro Hackers hide ransomware tools inside virtual machines using QEMU, allowing attacks to bypass security controls and remain largely invisible. …