Hyderabad Police Commissioner V.C. Sajjanar has expressed worry about the risks associated with ‘Agentic AI’ and emphasized the need for security measures and governance to address this issue. He pointed out that autonomous, multi-agentic Artificial Intelligence systems are now operating independently in crucial sectors, posing threats from both malicious exploitation and unintended behavior. Sajjanar highlighted the evolution of AI from basic ‘chatbots’ to ‘Agentic AI’ capable of decision-making and task execution without human intervention.
The Commissioner underlined the infiltration of autonomous robot agents into sectors like banks, hospitals, and power grids, raising concerns about losing control over these digital agents. He explained that Agentic AI goes beyond providing information; it can take actions based on its decisions, such as adjusting AC temperature or altering medication dosages. These agents collaborate with other AI systems to accomplish tasks, presenting a significant challenge due to their autonomy.
Sajjanar warned about the potential dangers posed by AI agents making thousands of decisions rapidly without human oversight, emphasizing the catastrophic consequences of even minor errors. He highlighted the risks of financial losses in the stock market, cybercriminal manipulation, and ‘Unintended Behavior’ where agents misunderstand tasks. The Commissioner stressed the importance of maintaining human control over these advanced agents and proposed the implementation of precise digital identities, logging movements, and establishing trust frameworks for human permission in crucial decisions.
