Context
CERT-In has issued specific auditing guidelines for autonomous “Agentic AI” systems capable of independent task execution, external tool utilization, and inter-AI delegation.
About Agentic AI
- Agentic AI refers to artificial intelligence systems capable of autonomously pursuing specific goals by planning, making decisions and executing tasks with minimal human intervention.
- Unlike traditional AI systems that primarily respond to user inputs, Agentic AI can independently determine the steps required to achieve a given objective.
- It combines reasoning, planning, memory and tool utilisation to perform complex, multi-step tasks.
Core Characteristics of Agentic AI
- Autonomy: Operates independently to achieve predefined goals with minimal human supervision.
- Goal-Oriented Behaviour: Works towards specific objectives rather than merely responding to individual prompts.
- Reasoning and Planning: Breaks complex tasks into smaller steps and determines suitable actions to achieve the desired outcome.
- Perception: Collects and interprets information from its environment, including text, images, databases and external inputs.
- Memory: Retains relevant information from previous interactions to maintain context and improve decision-making.
Applications of Agentic AI
- Healthcare: Supports clinical workflows, medical research and administrative tasks.
- Finance: Assists in financial analysis, fraud detection and risk monitoring.
- Education: Enables personalised learning and automated academic assistance.
- Cybersecurity: Identifies potential threats and supports incident response.
Specific Implications for India
- Demographic Shift: Catalyzes the transition of the workforce toward higher-value services, mitigating the heavy 40% employment dependency on agriculture.
- MSME Growth: Pilot programs indicate 15–20% productivity gains and 20–30% cost reductions for Indian MSMEs.
- Agricultural Extension: Enhances crop planning and localized weather forecasting, building on recent Kharif 2025 AI-outreach programs reaching over 3.8 crore farmers.
| About CERT-In (Computer Emergency Response Team – India) • Statutory Status: Established in 2004 under Section 70B of the Information Technology (IT) Act, 2000. • Nodal Ministry: Functions under the administrative control of the Ministry of Electronics and Information Technology (MeitY). • Primary Mandate: Serves as the national nodal agency for responding to computer security incidents, protecting Indian cyberspace. • Core Functions: Responsible for the collection, analysis, and dissemination of information on cyber incidents, issuing security alerts, coordinating emergency response measures, and formulating auditing frameworks (such as the recent AI guidelines). |
| Important CTC From this article: CERT-In |
Q. With respect to Agentic AI and India's cybersecurity framework, consider the following statements:
1. CERT-In is a statutory body established under the Information Technology Act, 2000, operating under the Ministry of Home Affairs.
2. Agentic AI refers to autonomous systems that execute end-to-end workflows with limited continuous human supervision.
3. Reward hacking occurs when AI systems bypass safety constraints because their programmed reward signals heavily favor task completion over behavioral boundaries.
How many of the statements given above are correct?
(a) 1 and 2 Only
(b) 2 and 3 Only
(c) 1, 2 and 3
(d) 1 and 3 Only
Answer: (b) 2 and 3 Only
Explanation:
• Statement 1 is incorrect: CERT-In is a statutory body established under Section 70B of the Information Technology Act, 2000, but it strictly operates under the administrative control of the Ministry of Electronics and Information Technology (MeitY), not the Ministry of Home Affairs.
• Statement 2 is correct: Agentic AI operates autonomously with limited continuous human supervision, relying on adaptive reasoning, tool integration, and persistent memory to execute multi-step workflows.
• Statement 3 is correct: Reward hacking occurs when agentic systems exploit loopholes to complete a task, prioritizing the achievement of the programmed reward over maintaining intended safety and behavioral constraints.