Only 8% of UK online adults use an AI tool in connection with their telecoms services, even though more than half of the same population has interacted with generative AI in some form over the past year, according to Ofcom's latest Technology Tracker research.

The regulator reported that 53% of adult UK internet users had used a GenAI tool or an AI customer service chatbot in the previous 12 months. Within telecoms specifically, 6% used AI to resolve a service issue, 3% to compare prices, products or services, and 2% to lodge a complaint. The figures come from an Ofcom review examining how artificial intelligence is shaping customer experience in the phone, broadband and pay TV markets for consumers and small businesses.

Ofcom traces the shift to the last five years, during which retail providers have exploited rapid advances in generative AI to support more consumer interactions. AI systems themselves are not new to the sector: operators have used them for decades to automate and improve network management.

Consumers split on AI service

The research shows a sharply divided customer base. Many people recognise the benefits of AI in everyday life, and some prefer it to human interaction. Others worry they may lose the ability to speak to a person at all. That hesitation has not stopped telcos from moving: operators are generally applying guardrails to AI applications and restricting the functions they perform, a cautious deployment posture Ofcom documents in the review.

Despite the mixed attitudes, Ofcom expects the proportion of customers using AI to navigate telecoms markets could grow quickly.

The next phase, the regulator suggests, is agentic AI — systems capable of planning, coordinating and executing tasks with a high degree of autonomy. An AI agent could select the best broadband deal for a user and prepare the transaction for approval. Further out, AI intermediaries such as AI assistants could monitor customers' telecoms accounts and sign them up to better deals automatically.

Consumers are not yet sold. While some recognise the time and effort personal AI agents could save, most share concerns about risks to personal data and privacy, and most still believe negotiating with a human gets them the best deal. Ofcom also flags structural obstacles before such applications can be implemented successfully: robust verification and authorisation protocols, plus broader regulatory, physical and commercial concerns.

Regulator draws lines

To date, Ofcom has seen limited evidence that AI applications have systemically harmed telecoms customers, and it believes its existing rules protect consumers effectively. But the regulator has now clarified its expectations of telcos and others building AI applications for the sector.

First, it must be very clear to telecoms customers when they are interacting with a GenAI application. A GenAI-augmented chatbot, for example, should make it transparent from how it describes itself that it is powered by AI. Second, Ofcom said it would be concerned if communications providers' AI applications were to act as "gatekeepers" creating barriers to support channels that providers are required to make available to customers.

The intervention matters because UK communications providers are already contractually obliged to maintain certain support channels, and the regulator is signalling that AI must not become a mechanism for degrading them.

Cristina Luna-Esteban, Policy Director at Ofcom, framed the regulator's position in consumer-terms. "AI has the potential to make the consumer journey in the telecoms market easier, quicker and more personalised. It could help people compare services, negotiate contracts and get more for their money," she said. "We're working with telecoms firms to ensure a future where AI works in consumer's best interests and firms ensure human support remains available when they need it."

Ofcom said it will continue to engage proactively with industry, consumer groups and other stakeholders on AI innovation and on changing consumer attitudes and behaviour — a signal that the current transparency expectations could evolve as adoption grows beyond the 8% who use AI in their telecoms dealings today.