Signal vs. noise: How telcos talk big on AI while delivering little on transparency

In February 2025, Hatem Dowidar, Group CEO of e&, took the stage at the Mobile World Congress (MWC)—the biggest international trade show dedicated to the mobile communications industry—to present the story of what the company had been working on over the preceding few years. A video filled with celebrity cameos showed a world where e& services are omnipresent in daily life, from home automation and e-learning to payments and delivery.

Over the past five years, telecom companies have become increasingly vocal about their ambition to transform “ from telco to techco,” i.e., expanding their businesses to platforms and software-driven digital services. Telcos want to remind everyone that they can be more than the mere cables and antennas powering the networks we use every day to go online. For more than a decade, telecom operators have watched technology giants—primarily Google, Microsoft, Amazon, Meta—drive outrageous market growth and increasingly concentrated capital. The dominance of Big Tech, they argued, had constrained them from fully capitalizing on the digital economy.

To reclaim some competitive space, telcos chose the route of collective action. Across jurisdictions, they demanded their “ fair share” for facilitating the traffic flowing to social media websites and video streaming services. To “level the playing field,” they advocated for over-the-top (OTT) services to pay for the infrastructure their platforms relied on.

While telcos’ lobbying for cost sharing is far from a thing of the past, some telecommunications companies have shifted their narrative substantially. Rather than confronting platform giants, they seek to partner with them. Rather than confine the use of automation to mundane uses like predictive maintenance, they venture to explain why they are well positioned to revisit their business model and claim a larger piece of the pie controlled by the technology industry. The ongoing fixation on “artificial intelligence,” fueled by the new generation of Large Language Models (LLMs), is setting the terrain in their quest for relevance.

Dowidar acknowledged this new landscape during his keynote as he described Etisalat's rebranding into e&, emphasizing the need to offer hyper-personalized experiences to its customers. At the same MWC session, MTN's CEO, Ralph Mupita, stated that the future of connectivity is “really about the future of platforms.” Mupita envisioned that AI would open new revenue streams in the next 5 to 10 years, with the company already exploring the use of AI in agriculture, healthcare, e-learning, microfinance, and other operational or management-related applications.

Christel Heydemann, CEO of French telecom giant Orange, echoed these sentiments a year later in an MWC keynote tellingly titled “Leading the Future: Intelligent, Inclusive, Unstoppable.” Heydemann reminded the industry that, in the “age of AI,” telecom operators have a “historic role to play” as omnipresent and essential conduits of the digital economy. Heydemann argued that while technology is accelerating, trust is not, and telcos are best positioned to become “architects of a trusted environment” for the development and deployment of AI.

As telecommunications companies build out the narrative of AI as a new dawn for the sector, we ask whether telcos have the right corporate governance processes and policies in place to match their grand visions.

The current landscape of telcos’ AI and algorithmic systems transparency

The term “artificial intelligence” does not have a widely agreed upon definition. Because of this, the actual technologies being developed and deployed end up obscured behind overbroad marketing terms. RDR’s current methodology includes 36 questions (elements) across 10 indicators that together assess the policies that govern companies’ algorithmic systems. We define algorithmic systems as systems that “use algorithms, machine learning and/or related technologies to automate, optimize and/or personalize decision-making processes.”

The results are alarming. Six years since we began tracking it, the average algorithmic transparency score among telecom giants stands at 15%.[1] Our findings reveal a telecommunications industry that affirms it is taking the necessary steps to use technology “responsibly” and "ethically" while largely failing to translate those commitments into transparent policies and processes grounded in international human rights.

Only half of the companies assessed—América Móvil, Deutsche Telekom (DT), MTN, Telefónica, Telenor, and Vodafone—explicitly commit to human rights in the development and use of algorithmic systems. Another four—AT&T, Axiata, e&, and Orange—disclose a commitment to “AI ethics” or principles that broadly mention privacy, non-discrimination, transparency, accountability, or fairness, but are not grounded in international human rights frameworks. The two remaining telcos, Airtel and Ooredoo, did not disclose any type of human rights commitment in the context of algorithmic systems or “AI.”

Even the strongest principles do not replace the need for thorough and clearly articulated operational policies that translate intent into practice. The interpretation of terms like “ethics,” “values,” and “responsibility” varies considerably across social, cultural, and economic contexts.

The ways in which telecommunications companies are deploying “artificial intelligence” vary from operator to operator, but the most prominent use cases are similar across the sector. Over the past three years, what technology companies have touted as “AI” mostly refers to the Large Language Models (LLMs) that power generative AI services. But when it comes to telcos, the current discourse mostly focuses on internal applications of LLMs to support operational efficiency, network optimization, and customer experience. By 2023, AT&T was integrating a custom generative AI tool across the company to “support and empower employees,” while e& was experimenting with using OpenAI’s ChatGPT for customer service.

However, though most of the world depends on telecom services for everyday communication, their potential risks and harms are not necessarily evident to the people using them. Eventually, those risks filter into “AI” systems and multiply through them if poorly governed.

Severe privacy violations are one such risk. Telcos amass vast amounts of first-party data—information they collect directly from their users. This includes phone call records, mobile data consumption, geographic location (companies are able to track past and real-time physical movement using cell tower triangulation), app usage behavior, web browsing history, SMS messages, and roaming history, among others. The scale of this collection alone should make customers question how companies protect personal data from data breaches and how it could be abused for advertising. As telcos’ ambitions to implement “AI” grow in various directions, we should expect companies to demonstrate a high standard of transparency to earn the trust they often claim to already have.

However, since 2022, telcos have made zero progress disclosing how their customers can control the use of their data to develop algorithms. In the 2026 RDR Index, Airtel was the only one to clearly disclose that it uses data from its users to train algorithms. Airtel's Privacy Center states that the operator “use[s] machine learning to enhance services quality, prevent fraud, and provide smarter solutions”. In contrast to Airtel, DT's AI Guidelines state that the company “always ha[s] the customer’s trust in mind” and that it is “obvious to [its] customers that they are interacting with an AI when they do.” But neither the guidelines nor any other policy appears to spell out whether DT uses customer data to develop algorithms.

Similarly, the privacy policy of Celcom—a brand under CelcomDigi, in which Malaysian telecom giant Axiata holds a 33.1% stake—explains that the operator may use personal data for “product development purposes, such as [to] explore and develop interesting new products and services”, but it does not provide any details on what that actually entails. Ooredoo's privacy policy states the company uses "artificial intelligence (AI) technologies" to "[a]utomate and improve customer service interactions (e.g., chatbots, voice assistants)," but there is no clarification about the existence of user controls.

The symbiotic relationship with Big Tech

But telcos’ own policy gaps on AI and algorithms are not the only source of the problem. Since our previous assessment in 2022, telcos have ramped up their efforts to partner with technology companies to integrate “AI” into their operations and services.

Much of the information that telecom operators share publicly is focused on how they are implementing AI internally and reshaping their strategy to grow their business-to-business services. But telcos’ steady stream of new collaborations with tech giants show an eagerness to stay relevant in a sea of uncertainty about whether the tech industry's push for AI will eventually pay off.

Worryingly, telco giants are acting as accelerators of “generative AI” adoption while also deepening integrations with Big Tech cloud services. In 2024, Vodafone and Google revealed a 10-year expansion of an existing partnership. Its aim is to leverage the British telecom firm’s 300 million subscribers across 50 countries in Europe and Africa to grow the use of Google Cloud and Gemini, Google’s flagship AI chatbot. In 2025, Airtel and Deutsche Telekom signed deals with Perplexity, offering a year of the Pro version of the company's generative AI service. Later that year, DT also announced a similar multi-year deal with OpenAI to “design simple, personal, and multilingual AI experiences across communication and everyday productivity”.

The language DT used in these announcements mirrors similar examples across the sector. It conveys an ambition to allow “everyone to be able to get to know and experience the advantages of artificial intelligence.” Like many of its peers, DT launched an ad campaign in conjunction with its deal with Perplexity encouraging people to ask any question in Perplexity AI's service.

These highly publicized moves further perpetuate the mysticism of “AI” as an omniscient and inevitable entity. But beyond the marketing headlines, quotes from press releases, and news reports, do telcos have the right governance processes and frameworks in place to address risks and harms?

Only half of the telcos we assess in the 2026 RDR Index share their processes to identify impacts to fundamental rights related to the development and use of algorithmic systems. Still, there have been some improvements. América Móvil introduced a clause in its privacy policy committing to conducting Data Protection Impact Assessments when “processing data using AI.” e&'s 2024 Annual Report explains that the company's Responsible AI principles "[m]andate that the AI system development process and usage follow current privacy and data protection rules while processing data to meet the highest standards of quality and integrity." Nonetheless, these are still isolated and non-committal examples in a landscape of near-total opacity about algorithmic human rights due diligence.

In 2024, Axiata launched a new group-wide program that aims to accelerate the company's transformation into a “TelcoTechco.” The company claimed to have scaled up the implementation of “Classical AI,” listing use cases such as fraud detection and dynamic pricing, while also advertising its progress in the use of “GenAI,” such as customer service chatbots and content generation for marketing. Thomas Hundt, Axiata's Group Chief Strategy & Technology Officer, stated that one of the challenges the company faces to become an “AI-native telco” is to “truly build a unified data ocean that brings together all types of data—network data, customer data, telemetry, and more—that a telco enterprise generates.” According to Hundt, “[o]nly then can monetization opportunities emerge—for example, through hyper-personalization.”

One of the most common AI use cases seen in telcos who operate a fintech business is analyzing data for credit-scoring models. Airtel offers an “Airtel Personal Loan” and explains how lenders use AI; Axiata operates a “full spectrum fintech arm” called Boost. But neither company has a comprehensive policy in place explaining, clearly and transparently, how it uses and develops algorithmic systems.

These companies are not alone. The majority of the telcos we assess failed to disclose an algorithmic system use policy and an algorithmic development and testing policy. We found only three exceptions. Telefónica's detailed AI Principles and Transparency Center explain how the company uses and develops algorithmic systems; Telenor's Security AI White Paper covers the company's governance processes; and Deutsche Telekom's professional ethics document for AI Engineering and Usage explains the process that employees should follow prior to developing new AI functionalities, during development, and after launch.

For the most part, the rest of the telcos offer high-level documents laying out their commitments and principles. These declarations are usually brief and use broad, superficial language, without providing detailed descriptions of the internal rules and processes. AT&T, e&, MTN, Orange, and Vodafone are all emblematic of this.

The road ahead

There are facets of “AI,” its impact, and its rapid, bubble-like proliferation that go beyond the scope of the RDR Index. This includes the labor of data workers and the environmental impact connected to the expansion of specialized data centers.

The latter is particularly relevant for telcos. Telecom operators are prime candidates to facilitate new data center construction, leveraging their extensive experience building core communication infrastructure. Consulting giants have encouraged this push with visions of new telco revenue streams worth tens of billions of US dollars.

Many are racing to capitalize on their ownership of infrastructure through lucrative business deals under the banner of “AI sovereignty.” Yet these partnerships overwhelmingly rely on the same tech giants whose dominance drives the world’s dependency on their products. For instance, by mid-2025, NVIDIA had announced plans for 18 new telco-led “AI factories,” a marketing term that encompasses the buildout of high-end edge computing architecture, hardware accelerators, and AI data centers. In 2026, Telenor established a “ sovereign cloud company” and Norway’s first “AI factory” project with NVIDIA and Capgemini, explicitly citing the need to escape “geopolitical uncertainty and growing dependence on global hyperscalers.” In India, a subsidiary of Airtel, partnered with Google under the advertising giant's $15 billion plan to establish an AI Hub in the country.

There are important questions about the potential longevity of the AI industry in its current form, replete as it is with delirious enthusiasm, questionable paths to profit, and moments of reality-distorting hype. But we must not lose sight of these technologies’ current and potential impacts on the people that they are supposed to serve. As gatekeepers of information flows, telcos’ automated network filtering can be abused to block websites and censor dissenting voices. Their immense databases can enable invasive profiling, tracking, and facilitate government surveillance. The credit scoring deployed in their fintech services may leave people out of financial options and deepen inequality. AI-driven customer service systems may disadvantage or lock out the elderly and people with disabilities.

Telcos say they are measuring their steps carefully as they venture further into the AI market. However, as of today, their policies are not up to par with their claims. If the giants of the telecom industry aim to earn the trust of their customers, they need to start by acting transparently in every phase of their adoption and deployment of technology, everywhere they operate.