Google AI is moving deeper into AI-powered cybersecurity, with the company at I/O 2026 placing security inside its larger enterprise AI story by combining model reasoning, threat intelligence, cloud infrastructure, automated vulnerability research and code repair. This timing matters because Anthropic’s Claude Mythos Preview, announced on April 7, 2026, changed the conversation around AI security; Anthropic chose not to release it publicly due to its advanced cybersecurity capabilities, instead placing it inside Project Glasswing, a controlled defensive programme. Mythos was described as available in private preview to select Google Cloud customers through Vertex AI. Google is now building its own defensive AI stack around Gemini, Big Sleep, CodeMender and its existing threat intelligence business, making it both a partner in the Mythos ecosystem and a direct contender. The strongest part of Google’s pitch is convenience, as bundling AI-driven capabilities into products enterprises already trust increases pressure on cybersecurity startups selling generic AI security products.