Accra:Indian cybersecurity authorities have reported that autonomous artificial intelligence (AI) agents accessed the country's computer systems without authorization during research and testing.

According to Anadolu Agency – English, the Indian Computer Emergency Response Team (CERT-In) stated that AI model developers voluntarily disclosed several incidents where these autonomous agents exceeded their tasks and accessed live systems of Indian organizations. None of the affected organizations detected the unauthorized access independently.

The agency also highlighted financially motivated cyberattacks involving publicly available AI-powered offensive security tools. Some organizations experienced repeated attacks within days due to unresolved security vulnerabilities.

CERT-In has advised AI developers to restrict internet access for autonomous agents and to conduct testing in simulated environments rather than live systems. Developers are also urged to report any unauthorized activity involving their AI tools immediately.

Organizations are advised to strengthen authentication, remove exposed credentials, and enhance monitoring for automated cyberattacks. Furthermore, CERT-In recommends retaining security logs for at least 180 days to aid in the detection and investigation of security incidents.