Data must be treated as a governed product: discoverable, structured and subject to clear contracts for ownership and use, so that AI agents can access it safely and consistently without manual intervention at every step. Agentic systems also need explicit security frameworks: common policy languages and guardrails that define authorized behavior, enforce boundaries, and generate machine-readable audit trails at scale.

Regulation is hardening this agenda into a requirement. The EU AI Act requires high-risk AI systems to meet audit and governance standards, and proposed European legislation would introduce enforceable sovereignty assurance levels for public sector AI procurement. McKinsey's 2026 AI Trust Maturity Survey reinforces the picture: organizations with explicit accountability for AI trust achieve consistently higher maturity scores, and investment in responsible AI is strongly associated with realized business value. Trust, the research concludes, is increasingly a business enabler rather than a compliance exercise.

Across the industry, operators and their partners are already building the collaborative, standards-based foundations this requires. The direction is clear. But individual effort will not close the gap at the speed or scale the industry needs.

Trust as an operational capability

Historically, trust has often been treated as a soft concept. Organizations make claims about being trustworthy. Vendors describe trusted platforms. Leaders communicate their commitment to responsible AI. Those discussions remain important, but they are no longer sufficient. In an autonomous environment, trust must become measurable.

Organizations need to demonstrate why an AI system made a particular decision, with full visibility into the data and models involved. They need evidence that controls function as intended and confidence that governance mechanisms hold as systems evolve. Trust therefore becomes an operational capability rather than a marketing message.

This mirrors previous advances within telecom: reliability, security and network performance all became measurable disciplines, not aspirational qualities. AI trust is following a similar trajectory. The future will belong not to organizations that simply claim their AI is trustworthy, but to those that can demonstrate it.

The road to autonomous telecoms

The telecommunications industry is moving steadily toward greater autonomy. AI agents will become more capable, models more specialized, and autonomous systems will assume increasing responsibility across networks, IT environments, and customer operations. The complexity of managing these environments will grow with them.

This creates an important paradox. The more AI organizations deploy, the more important governance becomes. The more decisions machines make, the more confidence leaders require in the systems making them. Autonomy and trust are not separate objectives. They are inseparable.

The industry often frames autonomy as a technology challenge. The real challenge is simpler and more fundamental. Can organizations understand what AI they have? Can they govern it effectively? Can they explain its behavior? Can they trust it to operate within acceptable boundaries? The operators that answer those questions successfully will be best positioned to unlock the full value of autonomous operations.

The race to autonomous telecoms is therefore not fundamentally an AI race. It is a trust race. And it starts with the basics: knowing what AI exists across the business, how it is being used, what data it relies on, and whether its decisions can be explained, governed, and trusted. Without that foundation, autonomy cannot scale safely.