Hyderabad, Feb 6 (NationPress) The Police Commissioner of Hyderabad, V.C. Sajjanar, has expressed serious concerns regarding the threats posed by ‘Agentic AI’, urging the implementation of suitable security measures, frameworks of trust, and governance to address these challenges.

In a recent social media statement, he emphasized the independent operations of autonomous, multi-agentic Artificial Intelligence (AI) systems across vital sectors. He pointed out that the dangers stem not only from malicious misuse but also from unintended autonomous actions.

The IPS officer noted that the landscape of AI has dramatically evolved. Once confined to simple 'chatbots' that only responded to inquiries or generated text, this technology has progressed to ‘Agentic AI’—systems that can autonomously make decisions and perform tasks.

“Autonomous robotic agents have infiltrated critical domains like banking, healthcare, and energy management. With these digital entities executing tasks without human oversight, there is considerable anxiety about losing control over their actions,” he stated.

The Commissioner highlighted the importance of public understanding of Agentic AI. For instance, while a conventional AI might answer a weather query, Agentic AI goes further by recognizing adverse weather conditions, closing windows, and adjusting air conditioning automatically. This indicates a shift from mere thought to actionable outcomes.

“In the financial sector, these agents can freeze accounts in response to suspicious activities, adjust medication dosages in hospitals based on patient conditions, and manage machinery in industrial settings—all autonomously. Their ability to communicate and collaborate with other AI agents to accomplish tasks poses a significant challenge,” he cautioned.

Sajjanar pointed out that these agents, capable of making thousands of decisions within minutes without human guidance, could create significant hazards if they err.

“For example, a single poor decision by an AI agent in trading could result in losses amounting to crores in an instant. There’s also the danger of cybercriminals manipulating these agents to engage in illicit activities. Moreover, there’s the risk of ‘Unintended Behavior’, where an agent misinterprets its assigned task and follows an incorrect path to achieve its objective. This is why experts insist that human oversight is essential for these advanced agents,” he explained.

The top cop believes that every AI agent should possess a clear ‘Digital Identity’. “Similar to how individuals have ID cards in an organization, these software agents must be identifiable. Tracking which agent opened which file, when modifications were made, and to whom information was sent must be documented (Logging).

“This way, if an incident occurs, we can swiftly determine which agent was responsible and correct the problem. Ultimately, stringent rules (Governance) are necessary regarding the autonomy of these agents and the data they can access. 'Trust Frameworks' must be established to ensure human authorization is obtained before critical decisions are executed. While accelerating tasks through technology is essential, ensuring security is paramount. Tech experts caution that failing to manage the pace of automation wisely could result in greater losses than the gains it provides,” he added.