TM Forum, Mauritius Telecom launch trusted AI model certification push

TM Forum and Mauritius Telecom are working to create what they say could become the telecom industry’s first standards-based certification framework for trusted AI model operations and monetisation. The initiative aims to close a trust gap as carriers move AI from experimentation into production.

Why it matters: - The telecom industry is moving AI models from pilots into production, but operators still struggle to prove those systems are governed, secure, observable and commercially accountable. - TM Forum research says 72% of CSP leaders are confident in the trustworthiness of AI in production, but only 14% can produce evidence to prove it. - A certification path for Model-as-a-Service, or MODaaS, could give carriers a common standard for trust, control and monetisation.

What happened: - TM Forum announced a global collaboration with Mauritius Telecom on trusted Model-as-a-Service for telecom AI operations. - Mauritius Telecom joined as a founding contributor and co-chair of the MODaaS collaboration. - The collaboration will define controls and conformance criteria for MODaaS across model lifecycle management, observability, security, trust, cost visibility and token-based monetisation. - The initiative is the second round of TM Forum’s MODaaS industry effort, known as GB1087.

The details: - The framework is designed to create a common operating model for sourcing, governing, consuming and measuring AI models as trusted enterprise services. - The work is meant to turn TM Forum’s MODaaS blueprint into a practical certification pathway. - The certification effort is expected to cover usage transparency, performance assurance, policy control, cost visibility and trusted token-based business models. - TM Forum said the completed framework is expected to become the world’s first standards-based certification framework for trusted AI model operations and monetisation in telecom. - Guy Lupo, TM Forum’s EVP for Trustworthy AI and Data Mission, said trust must be demonstrated and that the industry needs common mechanisms to prove governance, observability, security and commercial accountability. - Veemal Gungadin, CEO of Mauritius Telecom, said the next phase of AI in telecom will focus on operating, governing and monetising models in ways that customers, partners and regulators can trust. - Mauritius Telecom said its goal is to help shape a practical certification framework for the wider industry while showing how a sovereign AI platform can support trusted model discovery, transparent usage, performance assurance and token-based business models. - Mauritius Telecom is building on its sovereign AI programme, the Mauritius AI Registry and prior TM Forum Catalyst project work. - The Mauritius AI Registry provides a trusted discovery layer for sovereign AI resources, including models, agents and skills.

Between the lines: - Mauritius Telecom is using a national operator’s infrastructure and governance work to position itself as a template for broader telecom AI standards. - The focus on certification, rather than experimentation, signals a shift from AI capability to AI assurance and accountability. - Token-based monetisation suggests the industry is also trying to define how AI services will be priced and traded, not just how they will be controlled.

What's next: - TM Forum is opening the collaboration to CSPs, technology partners, hyperscalers, vendors and other ecosystem players. - Those participants are being invited to help shape the certification framework. - The work now moves toward defining the controls, trust mechanisms and conformance criteria needed for certification.