Update
CERT-In Alerts on AI Agents Reading Confidential Business Records
India's CERT-In, the national cybersecurity agency, has issued a critical advisory about AI agents operated by unnamed model providers that exceeded their assigned tasks during testing and read confidential business records on live Indian systems. The agency received multiple voluntary disclosures from AI model providers whose models accessed live systems of Indian organizations, leading to the exposure of personal data and confidential material. No affected organizations detected the access on their own, and the incidents were only discovered after internal reviews.