Telecom operators unprepared for AI safety regulations

According to the TM Forum, telecom operators lack the technical evidence to prove AI systems operate safely under impending new regulations.

Boards exhibit high tolerance for risk to secure public announcements detailing AI capabilities. However, the push toward an autonomous enterprise creates exposure for CSPs.

TM Forum CEO Nik Willetts used the opening keynote at DTW Ignite 2026 to identify trustworthiness as the top driver of telco brand value globally, outpacing both coverage and pricing metrics. Operators cannot prove their autonomous network flows remain safe at scale.

“Trustworthiness is the number one driver of telco brand value worldwide, ahead of coverage and price,” said Willetts. “Get trust right, anything is possible. Get trust wrong, and nothing else matters.”

In a research report conducted by TM Forum Insights and IBM’s Institute for Business Value, 130 decision-makers across global operators were surveyed. 72 percent of respondents claim confidence in their AI trustworthiness. However, only 14 percent of the total operators surveyed can produce externally reviewable evidence verifying that safety.

Hard deadlines for regulatory compliance

The EU AI Act forces high-risk AIe systems to meet audit and governance standards starting in August 2026. The proposed Cloud and AI Development Act introduces enforceable sovereignty assurance levels for public sector procurement.

Rakhee Chachra, Global Research Leader for Telecom and Media at IBM’s Institute for Business Value, states that regulation acts as a readiness accelerator. Regulatory demands increase the compliance burden but correlate directly with a stronger risk posture.

TM Forum announced the ‘Race to 2030’ at DTW Ignite, a plan to push operators toward becoming AI-native businesses built on autonomous networks and composable IT. Achieving this ambition requires evidence-based assurance.

Establishing a governance committee does not generate the continuous telemetry required to audit a live agentic AI system managing a private 5G slice. Governance determines the acceptable parameters of operation and provides the rules to monitor compliance. Assurance demands concrete proof that the systems obey those parameters in a live production environment.

Embedding continuous telemetry pipelines

Active monitoring platforms must capture decisions made by autonomous models concurrently. An automated routine rerouting traffic during a localised hardware failure provides a prime example.

The assurance system must log the exact parameters that triggered the rerouting decision, the sensor data analysed by the model, and the specific limitations enforced by the overarching safety policy. The prevailing compliance model relies heavily on periodic attestations from corporate auditors. This manual auditing approach fails completely when applied to autonomous systems processing thousands of network variables per second.

Architecture teams must embed compliance mechanisms within the software deployment pipelines to generate verifiable operational evidence. Engineers must design monitoring tools that analyse data continuously and generate operational evidence directly from the infrastructure.

The ideal solution mirrors cruise control in a vehicle, physically preventing the system from exceeding established parameters even if a fault occurs. Operators must build systems analogous to their existing revenue assurance platforms to monitor AI health.

Validation in production environments

TowerCos implementing predictive maintenance models for remote cellular sites illustrate the required architecture.

A central governance framework dictates that an autonomous dispatch system cannot route a maintenance crew to a remote tower during severe weather events. When the AI system goes live, it processes weather API data, crew location metrics, and tower sensor readings. Without an embedded assurance mechanism, the enterprise relies on blind trust that the algorithm interprets the severe weather rule correctly every time.

An effective assurance deployment requires a parallel verification pipeline. The autonomous dispatch system must output a cryptographic log of every decision path. The assurance pipeline simultaneously queries the identical weather API, runs a deterministic check against the established safety parameters, and records the validation. This creates the externally-reviewable evidence demanded by the impending AI safety regulations.

See also: Ericsson brings agentic AI to core networks

Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is part of TechEx and is co-located with other leading technology events including the IoT Tech Expo and Cyber Security & Cloud Expo. Click here for more information.

Telecoms is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.