With the use of artificial intelligence (AI) growing in the telecom sector, the Telecom Regulatory Authority of India (TRAI) on Friday proposed setting up committees at telecom companies to oversee its use, monitor its impact and ensure responsible deployment. According to the consultation draft floated by TRAI, telecom operators, including Reliance Jio, Bharti Airtel and Vodafone Idea, will have to establish an AI Governance and Safety Committee (AGSC) with the approval of their board of directors or an equivalent governing body.
The committee will review the performance of AI systems, identify potential risks and ensure corrective action wherever necessary. It will be headed by a senior executive and will review AI systems at least once every financial year.
Telecom companies are increasingly using AI for network management, customer services, fraud detection, cybersecurity and predicting network congestion. For instance, AI can predict a sudden increase in mobile data traffic in an area and help operators allocate additional network capacity. It can also detect faults in network equipment before they disrupt services.
However, TRAI said the incorrect functioning of AI systems could disrupt telecom services, affect consumers and threaten critical infrastructure.
Under the proposed guidelines, telecom operators will have to classify every AI system as high-risk, medium-risk or low-risk, depending on its importance to network operations and its potential impact on consumers. For example, an AI system that manages network traffic could be classified as high-risk because a wrong decision could disrupt mobile services in an area. On the other hand, an AI chatbot handling routine customer queries may fall under a lower-risk category.
Before deploying any AI system, companies will have to check whether it works correctly, produces fair results and can function reliably. Employees will also have to monitor these systems and intervene if they make incorrect decisions.
TRAI has also made it clear that telecom operators will be responsible for the outcomes of all AI systems used by them, even if they are purchased from third-party technology companies. For example, if an operator buys AI-based network management software from a technology vendor and it causes a network disruption, the operator cannot escape responsibility by blaming the supplier.
Companies will also have to maintain a record of every AI system they use, including those that are temporarily inactive or have been discontinued. They will have to obtain details of AI features built into network equipment and software from suppliers before purchasing them.
Further, operators will have to maintain a database of AI-related incidents, investigate the reasons behind failures and take steps to prevent them from happening again. Serious incidents, classified as critical or high severity, will have to be reported to TRAI.
An AI-related incident could include a system wrongly identifying a genuine customer as a fraudster and blocking the mobile number, or an AI system making an incorrect decision that disrupts network connectivity. Companies will also have to maintain records of consumer complaints arising from AI-related decisions and the action taken to resolve them.
The proposed guidelines aim to ensure that the growing use of AI in telecom networks does not compromise service quality, consumer interests, data privacy and network security.