Cisco Talos describes VOLTA-AUTODOXER stealing account data and asking AI models to assess victims. Whether any model returned a dossier is unknown.

Parameter says a flaw in ChatGPT shared conversations enabled code to access private chats; OpenAI fixed it Oct. 26 and rated the report P1.

A source analysis reports record CVE disclosures and a $6 billion Binance theft in October, while evidence of AI’s role remains mixed.

Anthropic’s multiagent tests show stronger collective vulnerability hunting—and coordination failures that complicate plans for always-on AI defense.

Berlin says its attack-detection system covered 5% of devices in one department. Rhysida stole and published about 1.4 million files.

Northflank says customers should assume environment variables were decrypted after a security incident; key services were listed as experiencing major outages.