Hyderabad: Hyderabad Police Commissioner V. C. Sajjanar warned that the rapid rise of autonomous “Agentic AI” systems posed serious risks, stressing that humans must retain firm control to prevent large-scale damage in critical sectors.
Sharing his views on X, Sajjanar said artificial intelligence had moved far beyond simple chatbots and text-based tools. He said the technology had now evolved into autonomous digital agents capable of taking decisions and executing actions independently. Already, such systems were operating in banking, healthcare, power grids, and industrial environments, raising concerns over reduced human oversight.
Unlike traditional AI, he explained, Agentic AI did not stop at providing information. Instead, it acted on its own decisions. For instance, it could freeze bank accounts after detecting suspicious transactions. Similarly, it could adjust medication dosages based on patient data. In industries, it could regulate machine performance. Moreover, these agents could coordinate with other AI systems to complete tasks.
Humans must retain control over autonomous AI systems
However, Sajjanar said this level of autonomy had created a major challenge. These agents could take thousands of decisions within minutes. Even a small error could trigger massive losses. In stock markets, he noted, a wrong AI decision could wipe out crores in seconds.
At the same time, he warned about cybercriminals hijacking AI agents for malicious purposes. In addition, he flagged the risk of “unintended behaviour,” where an AI system might misunderstand its objective and take a harmful path while trying to meet its goal.
To reduce these dangers, Sajjanar said every AI agent must have a clear digital identity. Authorities, he added, should log every action taken by an agent, including file access, data sharing, and system changes. As a result, agencies could quickly trace responsibility if something went wrong.
Finally, Sajjanar underlined the need for strict governance rules. These rules, he said, must define how much freedom AI agents receive and what data they can access. He also called for strong trust frameworks that require human approval for critical decisions, warning that unchecked automation could cause more harm than benefit.
Digital Agents… Danger Bells!
The journey of Artificial Intelligence (AI) has taken a new turn. Until recently limited to ‘chatbots’ that merely answered our questions or wrote text for us, this technology has now evolved to the level of ‘Agentic AI’—systems capable of taking… pic.twitter.com/GxLj0yCe8s
— V.C. Sajjanar, IPS (@SajjanarVC) February 6, 2026