Google unveils Gemini 4 Carbon AI model with advanced cybersecurity capabilities

Google◼ has announced Gemini 4 Carbon, a successor to Gemini 4 Argon, scoring 88.2% on SEC-bench Pro for vulnerability discovery, 54.8% on ExploitGym for exploitation and 71.0% on CWE-bench for patching, all evaluated without guardrails. On Gray Swan's Indirect Prompt Injection benchmark it recorded a 0.0% attack success rate versus Argon's 0.7%, which covers only the evaluated attacks. Access is limited to selected defenders via the Fairwind Program, so security teams should treat the results as vendor-reported and validate them independently.

For security analysts · Written by AI from 2 sources · Wrong?