A report from PwC revealed that AI poses trust and security (T&S) issues to telecom security, particularly in light of the rise of IoT and 5G.

Released this week, the inaugural Trust & Safety Outlook from the consulting giant reports AI introduces new vulnerabilities in telecommunications. In particular, as AI use in network performance has scaled to match IoT and 5G expansion, so too have increased attack surface and data exposure in the sector.

In an interview with SDxCentral, Dan Hays, PwC partner and telecom leader, said that telecom firms “should be worried about Trust & Safety with AI along multiple dimensions.”

The warning comes in spite of the still-nascent AI use case in telecoms. As Hays commented at Mobile World Congress Barcelona this year, the AI challenge is “overwhelmingly” defined by the lack of a business model that can trigger a new investment cycle from telecom operators.

With the report, the PwC partner encourages telecom firms to look beyond their investments, down to the end of the funnel, where AI introduces new avenues for misinformation, bias, and impersonation fraud.

“As potential enablers of AI, particularly through edge data centers, telecom operators have a vested interest in making sure that AI is safe and trustworthy, so that adoption can help drive their business model,” Hays said, adding that telecom operators should ensure the AI they use, as representing them to customers, is foolproof.

“Trust isn’t just a regulatory checkbox; it’s the foundation for adoption, and if telecom doesn’t bake it in from the start, any future AI gains may be compromised before they scale,” he added.

Hays also elaborated on the cybersecurity threats posed by increased AI proliferation. While AI-driven cybersecurity solutions are increasingly being used by firms such as Bell Canada and Globe Telecom, telecom firms cannot assume they are on a level pegging with the threat from bad actors.

With 5G splitting the network into “slices”, for example, an attacker can leverage AI to move from one slice to another, more critical one.

Here, Hays reminded firms to treat each slice in a zero-trust zone of its own by ensuring hardware-level virtualization controls with tightly segmented network functions. Constant real-time monitoring for anomalous behavior at the slice boundary is also recommended.

AI-powered IoT systems also face an increasing risk of device spoofing, deepfake signals, or malicious model manipulation, meaning telecom firms need to “shift from reactive safeguards to proactive, systemic oversight across their AI-IoT ecosystems”, said Hays.

A potential scenario is the leak of an AI model trained on an extensive data set - think IoT traffic data. The damage here would be on both a data privacy and business competition level, according to PwC.

“These models often encode insights into network behavior, customer patterns, and even infrastructure vulnerabilities; in the wrong hands, that’s a blueprint for disruption or replication.”

An easy safeguard is to isolate AI data from larger public models, with security installed from inception.

“Encrypt model data at rest and in transit, limit access tightly, and treat AI pipelines with the same rigor they apply to core network assets,” said Hays.

While the PwC report doesn’t highlight the nascent offering of 6G, its AI-native nature will likely pose trust and safety issues further down the road.

“Trust & Safety challenges won’t just be technical, they’ll be systemic and constant,” Hays agreed, with any vulnerability or misalignment in AI models able to “instantly ripple” across telecom services.

“To stay ahead, telecom firms will need real-time AI auditing, explainability baked into model design, and T&S frameworks that evolve as fast as the networks themselves.

“In 6G, trust won’t be an add-on, it’ll be an operating principle,” Hays concluded.