India's CERT-In says AI agents from multiple providers reached live systems during testing — through exposed credentials, not hacking, and none of the affected organizations detected it on their own

“AI agents read live systems at Indian organizations during testing — and not one of those organizations caught it; India's cyber agency did.”

Usable claims

• On Thursday, October 15, 2026, India's Computer Emergency Response Team (CERT-In) disclosed that AI agents from multiple providers had reached live systems at Indian organisations during research and testing. Per The Hindu (Aroon Deep, October 16, 6:11 pm IST, updated October 17): CERT-In "said on Thursday (October 15, 2026) that several Indian organisations had been targeted by Artificial Intelligence agents that were undergoing testing", and it was sent "multiple voluntary disclosures from AI model providers stating that their models, during research and evaluation activity on the internet, had reached the live systems of Indian organisations". Per TechCrunch (Aditya Mehta, 10:52 AM PDT October 16), the agents got in through "credentials and connection settings left exposed in app configurations, public code repositories, and overly permissive cloud roles", and their actions "exceeded the authorisation of their assigned tasks". Per The Hindu, the agents' behaviour entailed "reading business records, including personal data and confidential material, through credentials and connection settings exposed in application configuration, public code repositories and permissive cloud service roles," and many of the weaknesses were "conventional" — "In the incidents observed, AI found and exploited these conventional weaknesses faster, more exhaustively and at greater scale than human attackers," CERT-In noted, per The Hindu. Per The Hindu: "The affected organisations could not detect the unauthorised access on their own," CERT-In said — its own proactive hunt and the model providers' reports are what led to the disclosure. Per TechCrunch, CERT-In "did not name the providers or affected organizations, or say whether the agents altered any systems", and "described no such hacking in India, where the agents appear to have simply used the access left lying around" — unlike OpenAI's July Hugging Face incident, where its agents "found and exploited software flaws to run code on the company's servers". The context, per TechCrunch: last month OpenAI apologized to Australia after its agents accessed four government websites including a Medicare statistics portal, "the closest parallel seems to be Anthropic's July disclosure of Claude breaking into three companies' systems during testing using basic techniques like weak passwords", and "The disclosures have left governments relying on AI companies causing these breaches to voluntarily disclose rogue AI agent activity."