India’s telecom regulator on Friday issued a draft framework for the responsible use of artificial intelligence (AI), under which licensed entities under the Telegraph Act or the Telecommunications Act will have to be accountable for the outcomes of the AI systems they deploy, irrespective of their development origins.
The broad framework, issued for consultation, also mandates entities to adopt board-approved AI policies, establish dedicated governance and safety committees, and report serious AI-related incidents to the regulator, the Telecom Regulatory Authority of India (Trai) said in the paper, for which stakeholder comments have been sought until November 27.
The consultation on the framework comes at a time when AI is being increasingly integrated into networks to raise efficiency levels and into services being offered to consumers. Trai said that AI integration may also introduce new risks if due diligence is not exercised before the deployment of AI systems or if due supervision is not exercised over autonomous learning and systemic interactions.
“These risks, including inadvertent biases which may creep in AI systems during training and usage, unintended model behavior such as model drift, and adversarial attacks, can adversely affect the resilience of telecommunications infrastructure, particularly in its critical components. Such risks can be mitigated if suitable AI governance practices are followed,” it said while issuing a set of guidelines for consultation.
Indian telecom service providers, including Reliance Jio and Bharti Airtel, have been using AI within their systems at scale, most prominently for customer identification as well as fraud prevention. Global network equipment makers such as Nokia and Ericsson have been integrating AI into networks, which is also being adopted by Indian carriers.
Trai’s proposals include mandating each entity to formulate an internal AI policy, approved by its board of directors or equivalent governing body, covering risk classification, deployment approvals, incident management, human oversight, data governance and consumer grievance redressal. The regulator has also proposed that all AI systems be subject to appropriate human oversight for risk mitigation as per a predefined mechanism. Telcos and authorised entities using AI will have to classify their AI systems into high-, medium- and low-risk categories based on their significance to network operations, financial and operational outcomes, and potential implications for consumers.
Importantly, the proposals seek to make entities accountable for the outcomes of AI systems deployed in their networks and services, “irrespective of whether AI systems are developed internally, procured/ outsourced from a third party or a combination thereof.”
AI-related incidents would have to be classified into critical, high, moderate and low-severity categories, with mandatory reporting for critical and high-severity incidents. “Critical severity AI incidents and high severity AI incidents shall be informed to TRAI in a prescribed format, manner and timeframe,” it said. The regulator has not specified what will constitute critical, high or low severity or the duration within which mandatory reporting has to take place. Telecom companies will also have to maintain central databases of AI incidents and consumer complaints involving AI-related harm, with mechanisms for investigation, mitigation and resolution.
