For much of its history, telecommunications services have been inherently sovereign. Phone calls route through the network’s switching infrastructure, producing call detail records (CDRs) that fall under established policies and jurisdictional requirements.
Call control lives inside the network. Data stays put. The sovereignty gods are happy. But that long-time celestial tranquility is now being disrupted by the introduction of AI into the call path.
“AI voice is creating new threat vectors, including synthetic voice and real-time translation, that governments and enterprises are beginning to treat as sovereignty issues,” conclude the authors of the recently released Unthinkable Lab report The Telecoms Sovereignty Gap — How to Bridge It.
The moment AI is inserted into a voice call — transcribing it, analyzing it, routing it to downstream business workflows — a voice call stops being a simple, largely ephemeral event and morphs into something persistent and much more challenging to track. With AI, a single conversation, such as a three-minute healthcare screening, can produce multiple artifacts, including a recording, transcript, sentiment score, fraud risk assessment, updates to CRM or billing systems, and much more.
And each of those data-in-motion outputs comes with its own storage requirement, retention policy, and jurisdictional exposure.
For communication service providers (CSPs), the flourishing of the above-mentioned AI-powered data paths generated by voice conversations is creating new sovereignty and reliability risks that enterprise customers, especially those in highly regulated industries, are now raising in procurement conversations.
The Architectural Dimension
The first risk enterprises are now navigating is architectural. What happens if AI’s insertion into the call path becomes a potential point of failure?
CSPs sell voice services on the presumption of reliability and privacy. The network has redundancy, failover mechanisms, and SLAs measured in five-nines up time. A call should never drop because an AI transcription service hosted in a cloud environment three continents away misfires.
But that’s exactly what could happen if voice AI is integrated as a critical component in the call path. The CSP’s reliability and privacy guarantees now extend to a third party they don’t fully control. For regulated industries, including healthcare, financial services, and governments, third-party dependencies might be deal breakers.
The Jurisdictional Dimension
The second risk is legal. When data generated by a voice call enters an AI processing pipeline, it raises the question of whose laws govern that data.
Consider the scenario in which a UK healthcare provider calls a patient, and the CSP routes the audio to a US-based hyperscaler for transcription and clinical summarization. The conversation passes through a US company’s infrastructure, subject to the US CLOUD Act, which allows US authorities to compel that company to produce the data regardless of where the customer is located.
The UK healthcare provider fully expects doctor-patient conversations to be governed by UK law and NHS data-protection standards. Instead, it is now subject to US legal processes.
Before AI, voice call data paths were largely transparent. With the introduction of AI, data paths can become opaque. Lacking end-to-end visibility, CSPs have no idea if call data is being routed through multiple cloud zones, training datasets, or processing centers distributed across jurisdictions the CSP is unable to track or audit.
Keeping Tabs on Data
That’s why enterprise customers, especially regulated ones, are increasingly asking CSPs to document data paths as standard procurement practice. Questions they want answered include where the AI processing takes place, which company owns the AI model and can the CSP provide an audit trail. The inability to answer any of those questions is increasingly a disqualifier.
Standards bodies are now bringing some relief to the situation. The IETF, for example, introduced a common format for conversation records. Virtualized Conversations, known as vCons, offer a potential tool for addressing part of this problem.
By providing a structured format for conversation data, vCons point to the possibility of making data paths more transparent and trackable. If instructed, vCons can provide an audit trail showing which AI services processed which conversation and when. This potentially gives CSPs a mechanism for documenting which jurisdictions data traveled into and under what legal authorities it was processed.
More of a sovereignty-enabling tool than a sovereignty solution, vCons make the data governance story clearer and more auditable for customers and regulators. But they are only a small component of an effective voice AI sovereignty strategy.
A Tiered Approach
A more strategic approach to telecom sovereignty for CSPs is adopting a tiered solution that applies stricter data path controls based on reliability and privacy sensitivities. A tiered approach lets CSPs offer AI-enhanced voice to all customer segments while maintaining sovereignty guarantees where they matter most.
Calls involving regulated industries would occupy the top tier. AI processing for these conversations would run on infrastructure where the CSP has explicit jurisdictional guarantees and operational visibility.
Fortunately, that doesn’t mean restricting AI-enhanced voice calls to on-premises infrastructure. CSPs may also be able to work with regionally hosted platforms with contractual sovereignty commitments, or cloud partners that offer acceptable jurisdictional exposure and still meet sovereignty requirements.
Business and enterprise calls with fewer or no regulatory constraints would tolerate less-restrictive data path guarantees for AI processing, even those including hyperscalers. Regardless of sensitivity levels, CSPs will still need to document data paths and provide customers with audit trails.
For CSPs entering the voice AI space, the commercial case is straightforward: Tracking audit trails, classifying workloads and supplying vendor transparency will help put you in the strongest competitive position for attracting sovereignty-conscious business customers.
