Trai calls for human oversight, safeguards as AI reshapes telecom

COAI and UK government sign MoU to strengthen cooperation on AI, digital connectivity, digital trust and responsible technology-led innovation

India’s telecom regulator flagged the need for human oversight, safeguards and accountability as AI integration into telecom networks deepens network transformation, efficiency as well as threat detection.

 

“AI creates opportunity, but also requires safeguards through the right kind of governance approach,” Trai Chairman Anil Kumar Lahoti said on Monday, adding that appropriate governance of AI networks required their evaluation from the standpoint of public interest, consumer welfare and regulatory preparedness.

 

“An important principle that should govern this transition is that automation should improve outcomes, but not obscure accountability. Human oversight, auditability, and the ability to intervene where necessary remain important even as technology becomes more autonomous,” Lahoti added.

 

 

The sector regulator was speaking at the memorandum of understanding (MoU) signing between the UK government and the Cellular Operators Association of India (COAI), India's telecom industry body, to strengthen cooperation on AI, digital connectivity, digital trust and responsible technology-led innovation.

 

“Governments have a responsibility to ensure innovation progresses alongside security, accountability and resilience, and that’s even more essential when criminals are using AI to aid scams,” said Ben Mellor, Acting British High Commissioner to the Republic of India.

 

The Trai chief also flagged the dual role AI could play in combating online fraud and scams, saying the technology can strengthen trust by detecting anomalies and threats but, if misused, can also make harmful activities more sophisticated.

 

"AI can strengthen trust by improving detection of anomalies and threats, but if ill-utilised, it can also make harmful activity more sophisticated," he said, citing frauds and scams as clear examples.

 

AI can help identify suspicious patterns across datasets, but automated risk-classification systems need appropriate thresholds, safeguards and review mechanisms. He cautioned that wrongly identifying a legitimate consumer, enterprise or communication as harmful could lead to service disruption, reputational impact or denial of access.

 

"Fraudulent interaction can now move through multiple layers of the digital ecosystem. It may start with communication, involve a digital platform or an impersonated identity, and culminate in a financial transaction," he said.