The Next Frontier: Promise and perils of AI-enabled connectivity

Artificial intelligence (AI) is transforming how networks run and business is done, forcing a rethink of technology and policy alike. Self-optimising networks and predictive algorithms promise unseen efficiency, while AI-driven services and pricing models challenge old assumptions. Fully realising this potential requires updated principles that foster open competition, innovation and strong privacy safeguards. The sector must balance agility and financial sustainability with responsible oversight. The question is whether governance can adapt fast enough so that AI-fuelled innovation aligns with public interest, delivering inclusive and trustworthy connectivity.

Technological transformation

The deployment of AI can help telcos shift from being the proverbial “inactive pipes” towards full-fledged platforms with better experience for customers and enhanced revenues for operators. One of the major impacts of AI in telecom is transformation through network automation and predictive maintenance by offering real-time insights and anomaly detection. Vodafone uses AI to spot issues early, while Three UK applies Azure Operator Insights to unify data for actionable intelligence. AT&T is integrating AI across planning, spectrum and capacity forecasting for automating key tasks. Google’s Autonomous Network Operations, adopted by Vodafone and Deutsche Telekom, has reportedly helped cut repair times by 25 per cent. Parallelly, MTN Nigeria reduced outages by 20 per cent with AI monitoring, and NTT Docomo’s zero-touch pilot was able to free up 30 per cent of maintenance staff. These advances help improve reliability, lower costs and extend asset life.

The next major impact is AI-driven analytics, which are significantly impacting resource allocation and spectrum efficiency by monitoring conditions and reallocating capacity in real time. This is vital for 5G/6G with massive internet of things and low latency needs, enabling network slicing and adaptive bandwidth. T-Mobile’s pilot showed 15-20 per cent spectrum efficiency gains, extending rural coverage without major investment, while Verizon is advancing dynamic 5G slicing using analytics.

Another major impact of AI on the telecommunication sector is in enabling telecom operators to refine customer segmentation and deliver hyper-personalised experiences that drive growth and loyalty. Telefonica highlighted that AI deployment allows hyper-segmentation and predictive models to anticipate customer needs in real time, as implemented by Orange. The business impact of such innovations is certainly significant in terms of cutting churn and raising the average revenue per user (ARPU). Similarly, faster AI-driven services like automated chatbots reduce wait times and boost customer satisfaction and loyalty.

Cost equations

Telecom operators are set to cut expenses significantly through AI. Predictive maintenance can flag failures early, speed up fixes and extend asset life. The Boston Consulting Group notes that AI can automate complex order-to-cash workflows with end-to-end agents, reducing manual rework and driving major savings. AT&T is planning to save $6 billion, largely from network and maintenance, with 30 per cent fewer call centre contacts and optimised field dispatches. McKinsey projects generative AI and automation to increase earnings before interest, taxes, depreciation and amortisation margins by 3-4 points in two years, and up to 8-10 points in five.

AI innovation is helping telcos move beyond basic connectivity into new revenue streams and business models. Network slicing and premium quality of service tiers allow operators to sell specialised connectivity under strict service level agreements, as seen with SK Telecom’s slice marketplace or AT&T’s “Turbo” and Singtel’s “Express Pass.” Others are packaging AI as a service. Reliance Jio is partnering with Nvidia on GPU-powered clouds and its “JioBrain” platform to deliver affordable AI-as-a-service, alongside plans for an AI data centre.

AI is enabling dynamic, personalised pricing in telecoms. By analysing usage, congestion and preferences in real time, operators can adjust tariffs to encourage off-peak use, upsell bundles or tailor prepaid offers. Swisscom applies AI analytics to fine-tune individual plans. The benefits are higher ARPU, lower churn and efficient capacity use. However, it is important to note that without safeguards, opaque algorithms risk creating affordability gaps.

Telecom and AI: The public good imperative

AI strengthens telecom’s public-good role by expanding access to the 2.6 billion still offline and supporting sustainable development goals through better e-governance, health, education and finance. In rural areas, AI helps identify tower sites and uses drones for inspections, lowering costs and enabling coverage of remote villages. Each saving from AI efficiency can be reinvested to extend service to underserved communities.

Ensuring connectivity during disasters is vital for universal access. In the US, T-Mobile’s AI-driven self-organising network used real-time data to detect outages and reconfigure coverage by performing over 100,000 automated adjustments during recent hurricanes and wildfires. By analysing weather, social media and sensor data, AI can also provide early warnings, giving engineers and first responders crucial lead time to protect infrastructure.

As telecom networks become more AI-powered, security and privacy are critical. AI-driven tools detect fraud in real time, strengthen intrusion monitoring and block threats missed by rule-based systems. In 2025, Airtel Uganda launched Africa’s first AI-powered spam filter, protecting 11 million users. Such solutions cut SIM-swap and cyber fraud, building safer networks that foster digital inclusion.

Regulatory crossroads

The advances in automation, personalisation and efficiency illustrate the transformative power of AI in telecom. Yet, the very capabilities that make networks smarter and more agile also raise fresh regulatory questions.

One of the first issues to crop up is regarding net neutrality. AI-empowered network slicing and traffic prioritisation allow 5G networks to distinguish data streams with precision. For example, dedicating ultra-low latency slices to emergency services or telemedicine can be provisioned while reserving high-throughput channels for gaming or videoconferencing. But this sophistication can theoretically clash with the tenets of net neutrality. The fine-grained traffic differentiation by AI to help sustain more real-time bandwidth-hungry apps can mask favouritism or hide unwarranted discrimination under the opaque algorithmic management of data packets. It will be worthwhile to look at the existing frameworks in this aspect. In Europe, the Body of European Regulators for Electronic Communications rules allow AI-driven network slicing only if essential, without degrading general internet quality, and while keeping other traffic equal on a “best-effort” basis. In the US, the Federal Communications Commission bans paid “fast lanes”, though there are concerns in some quarters that slicing could breach non-discrimination rules. It is worth noting here that India has taken a balanced view, with the Telecom Regulatory Authority of India noting that services like telemedicine may justify optimisation, but not those that are substitutable over the open internet.

Second, a global framework for AI governance is still emerging, with divergent models. The European Union (EU) leads with a strict, rights-based AI Act that bans some uses and heavily regulates high-risk systems. The US takes a flexible but fragmented, sector-specific approach to avoid stifling innovation. The UK pursues a pro-innovation model, giving regulators non-binding principles while focusing on AI safety. China enforces swift, vertical regulations aligned with state priorities, aiming to balance public control with technological dominance. The Global South must also present a unified voice in shaping global AI frameworks.

Third, the debate on AI licensing often draws parallels with telecom regulation. Microsoft’s 2023 blueprint suggested licensing AI models and data centres like network operators. The global approaches differ, as the EU’s AI Act, 2024, uses a risk-based system without blanket licensing; the US emphasises testing and transparency over a new agency; while China requires generative AI platforms to register and obtain permits. So, it can be argued that licensing can ensure accountability through mandated oversight with advanced risk assessments, extensive pre-release testing and continuous monitoring of frontier models. However, it can be counter-argued that strict licensing risks can entrench incumbents and stifle innovation. A middle ground may be reached in terms of mandatory, auditable risk assessments for high-impact AI systems. There is broad agreement that accountability is essential and that regulation should remain flexible. In short, a responsible AI framework is the need of the hour.

The road ahead

Responsible AI (RAI) governance is a strategic necessity for telecoms. McKinsey estimates advanced RAI practices could unlock $250 billion by 2040, highlighting the need for ethics, transparency and accountability to secure trust and network safety. Yet, most carriers still apply AI ethics reactively, with few sector-specific standards. The GSMA’s new RAI Maturity Roadmap offers a self-assessment tool to guide ethical deployment, but implementing RAI demands model explainability, managing bias in training data, and “human-in-the-loop” for critical decisions. Going forward, collaboration among policymakers, regulators and industry bodies can establish shared models and oversight, helping telecoms build trust, ease regulation and future-proof innovation.

For telcos, the winning play is to go “AI-native” end-to-end with accountability at every touchpoint. For policymakers, it is becoming necessary to provide clear regulatory frameworks on issues where operators may face uncertainty, such as dynamic tariffs and slicing.  It is important for regulators to uphold open-internet principles and strike a balance between innovation, universal access and compliance. For civil society and customers, it is important to insist on transparency with clear disclosure norms. These steps build trust, reduce regulatory friction and keep innovation on track. The payoff of an AI-driven telecom network is immense, but it is equally important that the benefits are inclusive, fair and transparent.

In conclusion, the telecom sector’s adoption of AI is more than a technological upgrade – it is a redefinition of its public mission. From optimising spectrum and enhancing customer experiences to securing networks and extending coverage to the underserved, AI has the potential to make connectivity smarter, fairer and more resilient. The regulatory questions are complex but not insurmountable. If industries, policymakers and civil society move in sync, AI-powered telecom networks can ensure that the next era of connectivity is not only faster and cheaper, but also more trusted, equitable and sustainable – emerging as a cornerstone of inclusive digital economies and knowledge societies.

(The views expressed in this article are the personal views of the author)