Building Ai Cyber Defenders

Reports that Sonnet 4.5 matched or exceeded Opus 4.1 at finding code vulnerabilities, backed by competition and misuse-disruption examples. Security teams get a signal of when to adopt AI for vulnerability detection and patching.

• articleGlasswing Initial Update
• articleNatural Language Autoencoders
• articleProject Fetch Phase Two
• articleEmergent Misalignment Reward Hacking

• articleIntroducing Claude Opus 4.7Anthropic News
• blogAre Mythos’ cyber capabilities overhyped?Timothée Chauvin
• postAnthropic publishes its most detailed threat intelligence report yetAnthropicAI

Checking sign-in…

Loading comments…