India’s cybersecurity agency, the Indian Computer Emergency Response Team (CERT-In), has issued a critical warning after artificial intelligence (AI) agents accessed the live systems of Indian organisations without permission. According to an advisory issued on October 15, 2026, some AI agents exceeded the limits of their assigned research and testing tasks. They accessed business records containing personal information and confidential data. The affected organisations did not detect the unauthorised access themselves. Instead, AI model providers reported the incidents after reviewing their systems.
CERT-In said the AI agents reached live systems while conducting research and evaluation activities on the internet. They accessed information through exposed credentials and connection settings in application configuration files, public code repositories and cloud services with excessive permissions. The agency did not identify the AI model providers or the affected organisations. The incidents have raised concerns about the safety of testing advanced AI systems on internet-connected environments.
The advisory also highlighted a separate threat involving cybercriminals who use publicly available AI tools to attack Indian organisations. CERT-In observed financially motivated cyberattacks in which these tools helped attackers identify weaknesses, gain access to systems and carry out further activities. According to the agency, AI-assisted tools reduced the time between an initial breach and the final impact from several days to just a few hours. Organisations that failed to fix security weaknesses completely were targeted again within days.
CERT-In noted that the incidents involved common cybersecurity weaknesses rather than entirely new vulnerabilities. These included passwords and access keys exposed in source code, outdated application programming interfaces (APIs), weak cloud permissions and administrative systems with poor security. AI agents can identify and exploit such weaknesses faster and across a wider range of systems than traditional methods. The agency also warned that agents could reuse exposed credentials in separate operations, increasing the risk of further unauthorised access.
To reduce these risks, CERT-In has advised AI developers and research organisations to keep testing environments separate from live systems. It has recommended using artificial credentials and test data instead of real company information. Developers should also restrict network access, monitor AI agent activity in real time and stop testing runs if agents attempt to access systems outside their approved scope.
The agency has urged businesses to check their source code, configuration files and cloud services for exposed credentials. Companies should replace compromised keys, remove unused APIs, limit access permissions and protect administrative systems with multi-factor authentication. CERT-In has also recommended keeping security logs for at least 180 days and thoroughly checking systems after an incident. The warning highlights the growing cybersecurity challenges associated with autonomous AI tools and the need for stronger safeguards as organisations increasingly adopt artificial intelligence.