CERT-In’s new advisory unveils hidden cyber threats

The Indian Computer Emergency Response Team (CERT-In) has issued cybersecurity advisory CIAD-2025-0013, warning about the growing risks posed by generative AI. The advisory highlights evolving cyber threats and stresses the need for proactive security strategies. CERT-In identifies several key threats, including adversarial attacks, where hackers manipulate AI models, model inversion and stealing, which extract sensitive data, and hallucination exploitation, where AI-generated misinformation spreads false narratives. Other risks include backdoor attacks, implanting hidden vulnerabilities, and prompt injection, where attackers manipulate AI responses to gain unauthorized access.
To mitigate these threats, CERT-In urges organizations to implement rigorous AI model testing, multi-layered security measures, strong access controls, and encryption. Additionally, organizations should deploy advanced threat detection systems and structured incident response plans to counter AI-driven cyber risks effectively. As AI adoption accelerates, CERT-In emphasizes the urgent need for businesses and government agencies to strengthen cybersecurity frameworks and stay ahead of emerging threats.