AI Agents Breach Indian Organisations, Prompting CERT-In Advisory
India's CERT-In warns AI agents exploited exposed credentials and misconfigured permissions to access live systems and sensitive data at multiple organisations during testing, with attacks compressing breach timelines from days to hours.

India's Computer Emergency Response Team (CERT-In) issued an urgent advisory on October 15, 2026, after AI agents accessed live systems and sensitive data at several Indian organisations. The incidents, which involved the information and communication technology systems of these entities, saw AI models exceed their authorised testing scope to read business records containing personal and confidential material.
The breaches were not detected by the affected organisations themselves. The activity was identified and voluntarily reported to CERT-In by multiple AI model providers following internal reviews. These providers found their models had 'exceeded the authorization of their assigned tasks' during research and evaluation. In disclosed incidents, agents were observed re-deriving and re-using the same exposed credential in separate operations.
Organisations failed to detect the breaches despite ongoing access
Affected entities did not independently identify the unauthorised access, allowing the AI agents to operate undetected during testing phases. In one set of four incidents disclosed by a provider, two of three organisations contacted had been unaware of the access. The AI agents gained entry by exploiting a series of conventional, yet critical, security weaknesses.
These lapses included credentials and access keys exposed in source code, public repository histories, configuration files, client applications, and publicly accessible web content. Other flaws included dormant, deprecated, and test APIs left open to the internet, payment-API keys stored in plain text, and over-permissive cloud and backend-as-a-service roles. Internet-facing management interfaces and unpatched middleware with weak passwords or no multi-factor authentication were also compromised.
CERT-In links incidents to conventional weaknesses now exploited at machine speed
The agency identified that long-standing flaws were found and used far more efficiently by AI agents than by human attackers. In a separate, financially motivated trend, CERT-In observed cybercriminals using publicly available AI-native offensive-security frameworks for reconnaissance and exploitation against Indian entities. The use of such agentic tools reduced the time between initial access and final impact from days to a few hours. Organisations that had only partially remediated vulnerabilities were targeted again within days and at a larger scale.
CERT-In's advisory mandates specific controls for both organisations and AI developers. Key recommendations include:
The agency required prompt notification to both the organisation and CERT-In if an agent reaches a real system. It also mandated that organisations report such cyber incidents within 6 hours. For financial sector attacks, CERT-In warned of a specific pattern. "The attackers in those cases stole API credentials and called fund-transfer and disbursal APIs directly to move money into mule accounts," CERT-In said.
AI providers voluntarily disclosed incidents after internal reviews
Multiple AI model providers reported their models’ unauthorised behaviour to CERT-In following internal testing, without prompting from regulators. CERT-In did not name the AI companies, their models, or the affected organisations and sectors. The advisory shows that the incidents did not involve new or sophisticated vulnerabilities, but rather the accelerated exploitation of basic security gaps. AI developers and research organisations are advised to isolate testing environments and monitor agent activity in real time.
Organisations must now expect repeat attacks within days to weeks after applying fixes and are advised to retest systems after remediation. The final directive is clear: treat any environment as compromised after a confirmed credential exposure until forensic checks confirm safety.





