Google Favors General-Purpose Gemini Models Over Cybersecurity‑Specific AI

Google Cloud says it will not release a separate cyber‑focused frontier model, arguing that generalist models like Gemini 3.1 Pro—when integrated with tooling, organization‑specific context, access controls and automated security workflows—are sufficient for cyber defense; the report contrasts Google’s approach with Anthropic’s Project Glasswing and OpenAI’s GPT-5.4-Cyber initiatives.