India’s cybersecurity agency, CERT-In has issued a 'critical' advisory warning artificial intelligence (AI) agents have accessed live systems and confidential data of companies without authorisation. In its advisory titled ‘AI Agents as Attack Vectors’, the cyber agency said that threat actors are actively weaponising open-source agentic AI tools to execute rapid cyberattacks. It highlighted two distinct categories of incidents targeting the Information and Communication Technology (ICT) infrastructure of Indian entities.




CERT-In shares two incidents of AI models accessing company data without authorisationIn one case, AI agents accessed business records containing personal and confidential information. In another, cybercriminals used AI-powered tools to find security gaps and launch financially motivated attacks against Indian organisations.




During these incidents, the AI agents accessed business records containing personal data and confidential information. They used credentials and connection settings that had been left exposed in application configurations, public code repositories and cloud service roles with excessive permissions, the agency said.




Notably, none of the affected organisations detected the unauthorised access themselves, PTI reported.




AI tools cut cyberattack timelines from days to hoursAccording to CERT-In, the use of AI agents has sharply reduced the time between gaining initial access to a system and causing damage. In the campaign it identified, the gap shrank from several days to a few hours.




Organisations that only partly fixed the security weaknesses were attacked again within days, with the attacks carried out on a larger scale, the agency said. CERT-In noted that basic security failures were behind all the recent AI agent-driven incidents it observed. These included passwords, access keys and other credentials exposed in source code, public repositories, configuration files, client applications and publicly accessible web content.




Other weaknesses included old or test application programming interfaces (APIs) that remained accessible over the internet, payment API keys stored in plain text, cloud systems with excessive permissions, and management interfaces exposed to the internet.




Unpatched software, weak passwords and a lack of multi-factor authentication also left systems vulnerable. AI agents were able to find and exploit these gaps faster and on a larger scale than human attackers.




CERT-In's advice to companies and AI developersTo reduce these risks, CERT-In has asked organisations to check their source code, public repositories and their commit histories, configuration files, client applications and cloud storage for exposed credentials. The agency has recommended that companies change exposed credentials and remove them from code histories. They should also shut down unused, outdated and test APIs, limit cloud access to only what is required, take administrative interfaces off the public internet and enforce multi-factor authentication.




Organisations have also been advised to monitor systems for attacks and retain logs for at least 180 days.




CERT-In said companies should treat their systems as compromised after a credential leak or confirmed intrusion until a forensic investigation clears them. They should also prepare for repeat attack attempts within days or weeks.




For AI developers, research laboratories and evaluation agencies, the agency recommended isolating research and testing environments. Outbound network access should be blocked except where a task specifically requires it, with restrictions enforced at the network level rather than through instructions given to the AI model.




It also advised using synthetic targets and test credentials instead of real production data during research and evaluation.