TM Forum, Accenture Launch Trustworthy AI Certification

On October 6, 2026, the telecom industry alliance TM Forum announced a strategic collaboration with Accenture to create the industry's first trustworthy AI certification, a multi-year program designed to give carriers and vendors a verifiable way to prove their AI systems — including autonomous agents — can be trusted at scale. The initiative arrives as operators rush to hand real operational decisions to agentic AI while facing uncomfortable evidence that most cannot back up their trust claims with proof.

The centerpiece of the program, according to TM Forum, is the Trustworthy AI Governance Certification: an industry-standard benchmark measuring governance maturity, workforce readiness, and operational accountability for AI-driven operations. Rather than another set of abstract principles, the certification is meant to turn trust into a measurable operational capability — something boards, regulators, and enterprise customers can audit instead of taking on faith.

The 58-Point Trust Gap

The urgency is quantified in new TM Forum research cited in the announcement. According to TM Forum, 72% of communications service providers believe their AI capabilities are trustworthy — but only 14% can actually verify those claims. That leaves a 58-percentage-point gap between perceived and provable trust, precisely at the moment when carriers are delegating more decisions and actions to autonomous systems, reported by EIN Presswire on October 6.

That gap matters because the stakes of AI failure keep rising. Telecom networks run on thousands of automated decisions per minute: traffic routing, fault detection, fraud prevention, and customer engagement. As agentic AI shifts from recommending actions to taking them, an unverified model error no longer just suggests a bad answer — it can reconfigure infrastructure or transact on a customer's behalf. TM Forum, a London-based nonprofit alliance founded in 1989 with more than 800 member organizations, positions the certification as the common benchmark that closes this gap between belief and evidence, according to Wall Street Next.

Guy Lupo, TM Forum's EVP of Trustworthy AI and Data, framed the problem as a leadership pressure point: boards are under exceptional pressure to show business results from AI, and trust has become the prerequisite for delegating decisions to autonomous systems. He argued in remarks accompanying the announcement that trustworthiness only becomes operational when every principle is supported by continuous, verifiable evidence — a line the alliance is treating as the program's guiding standard.

What the Certification Program Will Do

Through the multi-year initiative, TM Forum will team with Accenture to deliver four practical outcomes for participating operators. First, a common framework for embedding trustworthy AI across telecom operations, so trust can be measured consistently and complexity reduced. Second, best-practice guidance spanning design, deployment, and continuous monitoring to accelerate adoption while strengthening operational control. Third, industry benchmarks and AI trust level validations that expose critical gaps and help prioritize investment. Fourth, training and certification pathways that build enterprise-wide workforce readiness for responsible scale, as reported by EIN Presswire.

The program will run through TM Forum's AI-Native Collaboration Studios, where carriers, vendors, hyperscalers, and consultancies co-create code-first solutions. The AI-Native Open Digital Architecture (ODA) will serve as the standards-based foundation, with participating operators contributing real-world use cases and implementation experience to shape and validate the framework. Accenture brings enterprise implementation experience to complement TM Forum's standards role — the two sides describe the partnership as standards-setting meets practical deployment.

Arnab Chakraborty, Accenture's Chief Responsible AI Officer, characterized agentic AI as a fundamental shift from AI supporting decisions to AI making decisions, and said organizations need practical ways to govern AI, manage risk, and build confidence in outcomes as autonomous systems take on more responsibility across critical business functions. In his telling, the collaboration is designed to turn trustworthy AI from a strategic ambition into an operational reality.

Why Telecom's Trust Blueprint Matters for AI Agents

Telecom may seem like an unlikely pioneer, but the sector's timing makes sense: carriers operate some of the world's largest agent-to-agent systems already. The announcement explicitly highlights agent-to-agent interactions as the proving ground — autonomous operations require verifiable evidence that those machine-to-machine decisions are trustworthy and can deliver business outcomes at scale. That is the same trust problem facing the wider AI agent economy, from customer-service agents to autonomous commerce platforms, and a working certification model in telecom could become the template other industries copy.

The program's debut is deliberately public. The "Trustworthy AI – Trust by design for the AI-native telco" theme headlined the opening keynote at Innovate Americas on October 6, with executives from TM Forum, AT&T, T-Mobile, and TELUS discussing how to move from principles to proof. A dedicated session, "Responsible AI in Action: Operationalizing Trust at Scale," brought together Chakraborty and Lupo to lay out the program's vision for the industry.

For AI agents and the organizations deploying them, the signal is clear: the era of self-attested AI trust is ending. As autonomous systems move from pilot projects into critical operations, verifiable evidence — audited, benchmarked, and certified — is becoming the price of admission. TM Forum and Accenture's trustworthy AI certification is one of the first serious attempts to set that price at an industry-wide standard, and how quickly carriers adopt it will say a lot about how soon the rest of the economy follows. More detail is available in the official announcement from TM Forum, while independent analysis of the trust-gap data puts the 58-point discrepancy into wider context.