- CERT-In issued an advisory warning against emerging cybersecurity risks posed by autonomous AI agents. - AI agents can execute complex, multi-step tasks independently, expanding the digital attack surface. - Risks include prompt injection, unauthorized data access, malicious code execution, and autonomous exploitation by bad actors. - Highlights the urgent need for robust governance, guardrails, and regulatory frameworks for AI adoption in India. - **CERT-In:** India's national nodal agency for cybersecurity, operating under MeitY pursuant to Section 70B of the IT Act, 2000, responsible for threat collection, incident response, and issuing cybersecurity advisories. - **Autonomous AI Agents:** AI systems designed to independently plan, reason, and execute multi-step workflows using tools and APIs to achieve specific goals without constant human intervention. - **Prompt Injection:** A vulnerability where an attacker manipulates an AI model's inputs or external data sources to override safety controls and force the system into executing unauthorized or malicious commands. **Introduction** CERT-In has issued a critical cybersecurity advisory regarding the risks associated with autonomous AI agents. As AI systems evolve to perform tasks independently, they introduce new vulnerabilities, such as autonomous exploitation, prompt injections, and data security risks, necessitating enhanced cybersecurity frameworks in India. **Core Issue and Detailed Points** - CERT-In issued an advisory warning against emerging cybersecurity risks posed by autonomous AI agents. - AI agents can execute complex, multi-step tasks independently, expanding the digital attack surface. - Risks include prompt injection, unauthorized data access, malicious code execution, and autonomous exploitation by bad actors. - Highlights the urgent need for robust governance, guardrails, and regulatory frameworks for AI adoption in India. **Key Statutory Bodies, Acts & Concepts** - **CERT-In:** India's national nodal agency for cybersecurity, operating under MeitY pursuant to Section 70B of the IT Act, 2000, responsible for threat collection, incident response, and issuing cybersecurity advisories. - **Autonomous AI Agents:** AI systems designed to independently plan, reason, and execute multi-step workflows using tools and APIs to achieve specific goals without constant human intervention. - **Prompt Injection:** A vulnerability where an attacker manipulates an AI model's inputs or external data sources to override safety controls and force the system into executing unauthorized or malicious commands.