AI can boost telecom security, but wrong decisions need quick intervention: TRAI chief

Artificial intelligence can transform the way telecom networks are managed and threats are detected, but its growing use must be accompanied by safeguards, human oversight and the ability to intervene when automated systems go wrong, Telecom Regulatory Authority of India (Trai) Chairman Anil Kumar Lahoti said on Monday.

Addressing the inaugural session of the industry body COAI’s ‘Powering India's Intelligent Connectivity Future’ event, Lahoti said AI was already helping telecom operators improve forecasting, optimise resources, identify network problems and respond more quickly to changes in demand.

He said the technology was paving the way for a shift from conventional automation towards networks with a greater degree of autonomy.

However, Lahoti cautioned that increased reliance on AI could create new risks if systems were not designed to deal with failures or incorrect decisions.

“AI creates opportunity, but also requires safeguards through the right kind of governance approach,” he said, stressing that automation should improve outcomes without obscuring accountability.

Lahoti said operators and technology companies should invest in responsible AI, conduct risk assessments, maintain decision logs and test models before deploying them. High-impact systems, he said, should continue to have human oversight.

“Human oversight, auditability and the ability to intervene will remain critical even as networks become increasingly autonomous,” he said.

He also warned that greater intelligence in telecom networks should not create new single points of failure. Systems should be designed so that incorrect automated decisions can be quickly overridden and essential services can continue if an AI-dependent function becomes unavailable.

“For consumers, trust is built through reliability, transparency, security, fair treatment, and the confidence that there is effective redressal when something goes wrong,” Lahoti said.

AI and online fraud

Lahoti highlighted the dual role AI could play in tackling online fraud and scams. While AI can detect anomalies and suspicious patterns across large datasets, he said the same technology could make harmful activities more sophisticated if misused.

“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.

He cautioned that automated risk-classification systems must operate with appropriate thresholds, safeguards and review mechanisms. Incorrectly classifying a legitimate consumer, enterprise or communication as harmful could result in service disruption, reputational damage or denial of access, he said.

“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,” Lahoti said.

He called for greater coordination between telecom operators, digital platforms, enterprises, financial institutions, government and law-enforcement agencies to tackle scams that increasingly cross different parts of the digital ecosystem.

“Consumer experiences the change as one event. Institutions often see only separate parts of it. That is why the response must increasingly be coordinated,” he said.

Lahoti said information sharing for risk management should be “purpose-specific, proportionate, and protected against misuse”.

Risk-based AI regulation

On AI regulation, Lahoti cautioned against both excessive prescription and leaving emerging technologies outside governance frameworks until problems emerge.

“A more durable approach is to focus on outcomes and risks,” he said, identifying consumer impact, accountability, security, reviewability of automated decisions and grievance-redress mechanisms as key considerations.

He also stressed the importance of technology neutrality, saying regulation should not lock the industry into a particular architecture or implementation model while the technology continues to evolve.

“Regulation should not lock the ecosystem into one architecture or implementation model, when the underlying technology is evolving rapidly,” Lahoti said.

At the same time, he said common standards and interoperability could improve security, reliability and coordination without restricting innovation.

Lahoti also pointed to the potential of AI-powered multilingual and voice-based assistance to strengthen consumer protection in India. Such systems could provide contextual warnings when users encounter suspicious interactions, particularly in a linguistically diverse country.

“India's next digital transformation will not be defined merely by how many people we connect, but by how intelligently, securely, inclusively, and responsibly we connect,” he said.