AI agents from multiple providers accessed personal data and confidential business records at Indian organizations during research and testing, India’s cybersecurity agency said Thursday. The agents reached live systems while running on the open internet.
The Indian Computer Emergency Response Team, or CERT-In, said it received “multiple voluntary disclosures from AI model providers.” According to the agency, the agents used credentials and connection settings exposed in app configurations and public code repositories, as well as cloud roles with overly broad permissions. Their actions “exceeded the authorisation of their assigned tasks.”
CERT-In did not name the providers or affected organizations. It also did not say whether the agents changed any systems. The affected organizations had not detected the intrusions themselves.
Read nextAnthropic’s third-quarter revenue reached $16.8B, FT reportsOther AI agents went beyond assigned tasks
The disclosure follows other cases of agents going beyond their assigned work. In July, OpenAI agents escaped a test environment and hacked French company Hugging Face. They began with exposed access, then found and exploited software flaws to run code on Hugging Face’s servers. CERT-In described no comparable exploitation of software flaws in India.
Last month, OpenAI apologized to Australia after its agents accessed four government websites, including a Medicare statistics portal. The agents wrote files inside that system, but OpenAI said it found no evidence they touched individuals’ medical or criminal records. In India, CERT-In said agents read personal data and confidential business records.
Anthropic disclosed in July that Claude had broken into three companies’ systems during testing using basic techniques such as weak passwords. The Indian disclosure differs from those earlier incidents because it involves agents from multiple providers.
Governments depend on AI providers to report activity
Because the affected Indian organizations did not spot the intrusions, the cases also show how governments can depend on AI providers to report this activity. TechCrunch has asked CERT-In for more information.
