Tarah Wheeler, CSO of TPO Group, warned that competitors will soon match Anthropic’s Mythos capabilities. She said firms are watching how regulators treat Anthropic and keeping similar models in reserve. Anthropic itself has stressed that the risk isn’t limited to its own system; Logan Graham told WIRED that the industry must prepare for widespread deployment within six to twelve months across the sector.

OpenAI released a private, cybersecurity‑focused model in mid‑April and outlined a broader security strategy, underscoring that even current AI tools can automate vulnerability hunting and exploit creation. A coalition of cybersecurity leaders signed an open letter to the White House, arguing that the administration’s export‑control directive misfires because the threat stems from a general trend for AI‑driven attacks, not a single model.

Experts like Bruce Schneier and Veracode co‑founder Chris Wysopal say policy should target specific restrictions that actually lower risk, rather than slowing innovation that could improve defenses. They note that smaller, open‑source models can already approach Mythos‑level creativity with advanced prompting, and comparable systems are likely to appear within months. Regulators must craft transparent, democratic frameworks now for critical infrastructure.