Google, Anthropic, and OpenAI Unveil Cyber AI Models, Safeguards, and Access Programs
Summary: Google announced Gemini 3.8 Flash Cyber, its most capable cybersecurity model, and made it available to trusted defenders through a new Fairwind Program that includes over 650 partners such as CrowdStrike, Datadog, Menlo Security, Palo Alto Networks, and Snowflake. The same week, Anthropic launched Claude Fable 5.1 and Claude Mythos 5.1 with tiered safeguards and new Enterprise Frontier Safeguards, while OpenAI said its forthcoming Astra model meets the Critical cybersecurity capability threshold under its Preparedness Framework and will reach testers through the Daybreak Blue program.
Key Takeaway: OpenAI reports Astra scored 100% on ExploitBench, declined 91.5% of jailbreak requests versus 59% for GPT-5.6 Sol, and found and chained two zero-day vulnerabilities during an evaluation - a model its own framework labels capable of a full attack on a hardened target from a high-level instruction alone. Anthropic's response to Claude models being used against real systems included pausing external cyber evaluations of pre-release models, a concession that the incidents were a failure of operational security.
Read the original at The Hacker News ↗
- Topics
- Cloud Security
- Related
- Google Cloud Warns Three AMD Turin CPU Flaws Could Leak Data Across Processes in VMs Cyber Risks · Nov 12, 2026New Falcon Cloud Security Innovations Secure the Cloud AI Runs On Cyber Risks · Nov 04, 2026