India’s premier cyber response unit, CERT-In, has issued a critical advisory highlighting the accelerated threat of cyberattacks powered by artificial intelligence (AI). The agency underscores that hackers are leveraging AI agents to drastically reduce the time required to inflict damage on compromised systems from days to just hours.
Emerging Threats from AI-Powered Attacks
On October 15, 2026, CERT-In released a detailed warning concerning AI-driven cyber threats targeting Indian entities. The advisory reveals two primary risks: the deliberate exploitation of open-source AI tools by cybercriminals to infiltrate systems, and unauthorized access by AI agents to operational business systems, as reported by AI service providers post-internal audits.
Financially motivated attackers are harnessing publicly accessible AI-based security testing frameworks. These frameworks enable agents to strategize tasks, execute commands, and persist online for prolonged periods, frequently using multiple agents to achieve their objectives within hours.
Security Vulnerabilities and Entry Points
The advisory identifies several common vulnerabilities, including exposed credentials in source codes, configuration files, and public repositories. Additional weaknesses stem from unutilized APIs, excessive cloud permissions, and inadequately secured management panels, often lacking robust authentication measures.
Automated attacks, as detailed in Cybersecuritynews.com’s report, emphasize the urgency for organizations to address high-risk vulnerabilities swiftly, with CERT-In advocating for remediation within 12 hours. The rapid nature of these attacks significantly reduces the window available for defenders to respond effectively.
Recommendations for Enhanced Security Measures
CERT-In advises organizations to regularly update and rotate exposed credentials, eliminate superfluous APIs, and enforce stringent cloud permission policies. Moreover, continuous monitoring of API request surges, bulk data access, and automated account activities is crucial, with log retention extending to a minimum of 180 days.
For AI developers, the advisory recommends default restrictions on outbound network access, allowing only essential connections, and employing synthetic credentials alongside disposable testing environments. These measures should be applied at the network level to prevent reliance solely on AI model instructions.
CERT-In’s comprehensive security blueprint for AI applications is designed to mitigate risks, emphasizing the need for regular external audits and forensic evaluations post-remediation. The agency warns that incomplete solutions leave systems vulnerable to recurrent, potentially more severe, attacks.
As the threat landscape evolves, keeping defenses updated against new malware and integrating threat intelligence feeds from extensive security operations centers is vital to maintaining robust cybersecurity.
