AT&T and GSMA warn Telco AI Models Drift

Boost Mobile data scientist Priyank Jain, GSMA director of AI technologies Louis Powell and AT&T data office vice president Mark Austin flagged a specific and seldom discussed risk in telco AI deployments this week: the gap between the data a model was trained on and the data it actually encounters once it is live, a failure mode known as train serve skew. Speaking to Fierce Network on 25 September 2026, Jain explained that the danger is not a crash or a failed job. Nothing alerts an operator that anything is wrong. Jain said the model "just gets quietly worse."

Two Different Ways Models Drift

The article identifies 2 separate causes. General purpose frontier models arrive with no exposure to telecom specific data formats or vendor taxonomies, since none of that appears in their internet scale training data. Telecom specific models face a subtler problem: production data, such as settlement records, arrives late or out of order in ways the training set never captured, and datasets in this domain routinely carry over 100 columns of custom, operator specific parameters. Powell added that frontier models have no inherent understanding of telecommunications and can produce confident, wrong answers, a real risk once a model is embedded in an enterprise workflow. Austin's prescription is procedural rather than technical: match training data to production sources as closely as possible, run a new model silently alongside the live system before cutting over, and keep testing it after deployment rather than treating the launch as the finish line.

Why This Matters Before Autonomy

This is a narrower and more mundane problem than the industry's current preoccupation with agentic AI and autonomous networks, and that is exactly why it matters. The NGMN report on agentic AI concluded that even coordination within a single operator's own domains, let alone accountability between an enterprise and its network provider, requires common information models and audit trails that do not yet exist. Train serve skew shows the problem starts a level below that. A supervised classifier with a known, static task can still fail silently once its inputs drift from what it was trained on. An autonomous agent negotiating across an operational boundary carries the same drift risk plus the coordination and accountability gap NGMN identified. Operators evaluating agentic AI pilots should ask a more basic question first: whether they would know if the model already in production today has quietly stopped working.

Quiet drifts like this example are worse than explicit failures. The quality of the output degrades silently, which degrades the quality of service. Explicit quality control for data quality before and after transformation, and before and after agentic intervention are necessary to enable autonomous networks.

Orange Business Moves Agentic AI Onto Open-Weight Models

Orange Business is moving its most sensitive AI workloads, including its agentic systems, onto open-weight models running on infrastructure it controls, and it says sovereignty rather than cost is the reason. The account comes from a Fierce Network interview with Miguel Alvarez, chief data and AI officer of Orange Business, published on 14 September by Mitch Wagner. Alvarez describes 3 changes in the operator's AI strategy over the past year: a shift of sensitive work from public cloud models to open-weight models it hosts itself, some of them Chinese; a move from chat windows beside applications to agents inside the systems of record; and a 3-month mandate given to a mixed team of AI and operations process experts to automate the front third of the incident management chain. That last piece is the one with a number attached. Qualification, triage and routing of a ticket used to take a technician 30 to 40 minutes working down a checklist of connectivity and configuration tests. Alvarez says an agent now does the same work in about 3 minutes, with 95% of it automated, and that the Level 1 support team "may not survive in its current form".

Sovereignty is the differentiator

The first change is the one Orange Business leads with. Alvarez says "the importance of being able to do AI in a trusted environment has increased a lot", that open-weight models now run work which needed frontier models 4 to 6 months ago, and that the models his teams use most for coding are Alibaba's Qwen and MiniMax, with Moonshot's Kimi under test. The sovereign option is also a product: Orange Business's Live Intelligence platform, built for 100,000 Orange Group employees and sold to 150 enterprise customers, serves Gemini, ChatGPT and Claude alongside Mistral and the open-weight models Orange hosts itself, so a customer can route ordinary work to a hyperscaler model and keep sensitive work inside a French enclave. Fierce reports that the optionality became a practical priority in June, when the US government ordered Anthropic to suspend foreign access to its 2 most advanced models. In July I argued that sovereign AI capacity is the most immediate of the credible operator revenue lines, and that edge and distributed compute become fundable when sovereignty is the demand driver rather than the garnish. Orange Business is the first operator I have seen describe the same logic from the inside, as a buyer of models rather than a seller of GPUs. It did not move to open-weight models because they were cheaper. It moved because a customer, or a regulator, or a foreign government, can now switch a frontier model off.

The second change is the one that matters for the business case. When Bain warned last week that agentic AI could raise telco opex by 30%, the mechanism was that a process augmented rather than replaced removes nothing: the human team keeps running the workflow while agents perform isolated tasks at its edges, and the legacy 70% to 80% of the cost base stays. Alvarez describes exactly that trap in his own first phase. Summarisation, ticket correlation and root cause analysis were the early work, and "you still have the same roles doing more or less the same work in more or less the same way". The 3-month mandate was the correction. It targeted a whole segment of the chain rather than a task inside it, it was staffed by process experts as well as AI experts, and it ends with a change to the organisation: Level 1 teams that supervise agents and handle customer communication instead of running the checklist. Similar exercises are running at Orange Group level across HR, finance, legal, software development and B2B sales, aimed, in Alvarez's words, at whole job lines rather than individual use cases. That is the difference between a demonstration and a P&L entry, and it is the first operator account that describes the retirement of the legacy process as the goal rather than as a consequence to be managed later.

The 2 changes are connected. The reason sensitive work moved onto controlled infrastructure is that agentic systems, "whose autonomy introduces risks requiring greater scrutiny", are now among the workloads. Fierce lists the season's incidents: an autonomous agent system breaking into part of Hugging Face's production infrastructure in July, a swarm of agents using a dormant wiki as a coordination board, a model reaching third-party systems during a security evaluation. Alvarez's response is operational rather than philosophical. Trusting vendors to watch on the operator's behalf is no longer sufficient; vulnerability assessments are more frequent; and the operator needs the ability to quarantine sensitive services and cut their connectivity. That is the containment layer I described when I argued that the agent runtime is not the agent model, and it is the enforcement half of the governance question the industry has mostly discussed as policy. An operator that runs the model on its own hardware can quarantine it. An operator that calls a frontier model through an API can only stop calling. Orange Business's stack, on Alvarez's description, is built on LangChain and LangGraph, OpenTelemetry for observability, Model Context Protocol for interconnection and open-source components for the LLM gateway and MCP registry. Every one of those is a runtime component. None of them is a governance model, and the interview does not describe one.

The main challenge for AI remains trust. Delivering trustable operation, results in a controlled, auditable, traceable manner is paramount. As a result, operating on models that are under your control, deciding what dataset should remain on premise, in jurisdiction or in public cloud and what governance model agentic relies on is a crucial set of decisions for operators.

Bell Canada Launches Self-Serve Network as a Service

Bell Canada announced on 10 September the Bell On-Demand Network, which it describes as the country's first network-as-a-service platform aligned with the 7 customer experience attributes defined by Mplify, the industry body formerly known as MEF: on-demand, observable, manageable, programmable, secure, modular and flexible. TelecomTV carried the announcement in its 10 September roundup. The platform runs on Bell's fibre network with Cisco 8000 Series Secure Routers underneath, and the operator's description is that business customers can, through a single self-serve platform, "autonomously order, activate, manage and scale connectivity". The first product is On-Demand Internet, available in Ontario and Québec wherever Bell has deployed fibre, with additional networking, security, cloud connectivity and automation capabilities to follow. Bell's CTO Mark McDonald said customers "can now provision, scale and adjust their network in real time, getting the exact bandwidth and services they need, the moment they need them". No price, no provisioning time, no API documentation and no customer figure were published.

The customer configures, the operator does not pre-package

I have argued for some time that the natural end state of differentiated connectivity is slicing as a service: a platform through which third parties discover, configure, reserve and consume network resources on demand, because enterprise CIOs know their connectivity needs better than the operator does and are used to configuring cloud services rather than selecting from a catalogue. Bell's launch is that model applied to fixed access, and it is worth noting what is different from the 3 mobile launches of the past 3 weeks. EE sells a priority lane as a tariff, Vodafone sells a predefined quality profile through an API, and Telstra sells a slice for 1 application chosen by the operator. In each of those the operator decides what the product is and the customer decides whether to buy it. In Bell's description the customer decides the bandwidth and the timing, and the operator's role is to make the change when asked. That is the inversion the slicing-as-a-service argument depends on, and Bell states it as the product rather than as a roadmap.

Fixed first, because bandwidth on fibre is a reservation

It is not an accident that the self-serve model arrives on fibre before it arrives on mobile. When Vodafone launched quality on demand I wrote that a priority is not a reservation: a mobile profile improves the customer's place in the queue on a congested cell but does not guarantee what comes out of it, and the profile parameters were not published. On a dedicated fibre access line a bandwidth change is a reservation. The capacity exists, the router policy is deterministic, and the operator can honour the customer's chosen figure without a proof-of-value engine to check whether it did. That is why Bell can let the customer set the number and why the mobile operators cannot yet. It is also why the interesting part of Bell's roadmap is the part that is not launched: cloud connectivity and security are where the customer's requirement starts to depend on a second network and a second party, and the on-demand model will be tested at the first boundary it has to cross. Mplify's 7 attributes describe the customer's experience inside 1 provider's platform. They do not describe what happens when the customer's agent asks 2 providers for the same thing.

Programmable is the attribute that matters

Of the 7 attributes, 6 describe a good portal. The seventh, programmable, is the one that decides whether this is a self-serve web page or a network resource that a customer's own systems can consume. Bell's announcement describes a platform, not an API, and does not say whether the ordering, scaling and observability functions are exposed to the customer's software or only to a person logged into a portal. The distinction is the one I drew at DTW Ignite: the APIs that ship are the ones that ask the network a question, and the ones that ask it to change its behaviour on a third party's instruction are harder. On-Demand Internet is the second kind, in the easiest domain. If it is programmable in the Mplify sense, a customer's cloud automation or an enterprise agent can scale the access line up before a backup window and down after it without a human on either side, and Bell has built a resource the customer's software can reason about. If it is a portal, Bell has built a better ordering experience. The announcement does not say which, and it is the first thing a CIO evaluating it should ask.

What has not been published

No price relative to a standard business fibre contract, no provisioning time for a bandwidth change, no statement of whether the platform is API-accessible or portal-only, no minimum term or change frequency, no customer count and no date for the cloud connectivity and security capabilities. 4 things to watch. First, published API documentation, which is the test of the programmable attribute. Second, the price of a bandwidth change against the price of a permanent upgrade, because the on-demand model only pays for the customer if variable capacity is priced below fixed capacity over the period it is used. Third, whether Bell extends the same self-serve model to its mobile network, where it would need 5G standalone slicing and a service level of the kind Telstra attached to its slice. Fourth, whether the cloud connectivity capability, when it arrives, lets the customer configure the far end as well as the access line, which is the first crossing of an administrative boundary and the point at which slicing as a service stops being a single-operator product.

Agentic AI Could Raise Telco Opex by 30%

Two reports published on 10 September describe the same balance sheet from opposite ends. Bain & Company, as reported by Telecoms.com's Mary Lennighan, warns that operators could see operating costs rise by around 30% without any gain in productivity or growth if agentic AI is bolted onto legacy operating models. The consultancy describes an emerging agentic operating model in which traditional costs account for 70% to 80% of an operator's total and AI costs for 20% to 30%, with the risk that the legacy 70% to 80% never goes away. The same morning, TelecomTV's James Pearce published a second cut of the AI-Native Telco Index, the 276-page evaluation of 56 operators based on published material from January to June 2026, and found that 20 of the 56 frame their AI strategy primarily around revenue, 28 balance revenue and efficiency, and only 6 are primarily focused on efficiency. TelecomTV names the 6 as KPN, Proximus, Spark New Zealand, BT Group, Comcast and Rogers. It also notes, in the same article, that across all 56 operators cost savings remain the most clearly measured AI-related financial impact.

A new cost layer on top of the old one

Bain's argument is about arithmetic rather than technology. Model prices have fallen almost 10 times in the past year, according to the report, but token usage is rising faster as staff find new uses in network operations and customer care. At roughly half the token price, a 4.5 times increase in usage doubles the bill. The recommendation is to measure the cost of a business outcome, a resolved customer issue or a closed network incident, rather than the cost per token, and to hold someone accountable for model drift and token spend. AT&T is the worked example: it redesigned its orchestration so that large supervising agents delegate to smaller specialised models, which by AT&T's own account cut costs by up to 90% while tripling throughput. Bain's 3 cost traps are the token bill, the dual operating model in which human teams keep running the process while AI performs isolated tasks at the edges, and the low-risk demonstration that improves efficiency at the margin without touching the P&L. Internal chatbots, call summaries and copilots are named in the third category. This is the cost side of the 3 money flows I set out in July, when I argued that cost avoidance and revenue must be separated before an operator can say what AI is worth to it. Bain adds the point that cost avoidance is not automatic either: the avoided cost has to be removed from the organisation, and a process that is augmented rather than replaced removes nothing.

Revenue is the stated priority, savings are the measured one

The Index finding runs the other way. When I covered the first edition of the Index on 7 September the observation was that none of its 4 dimensions was financial. The second cut is TelecomTV's answer, and it is a useful one, because it reads the same public record for money rather than for capability. Of the 20 revenue-first operators, 11 are Tier 2 operators with annual revenues between $2 billion and $10 billion, 11 are in Asia and 4 in the Middle East, which TelecomTV reads as evidence that the shift toward AI revenue is not being led from Europe or North America. The examples with numbers attached are few. Bell Canada reported around C$700 million of AI-powered revenue in 2025, up 60%, and targets around C$2 billion a year by 2028. SK Telecom's AI datacentre business grew 92.5% year on year to 136.2 billion won in the second quarter, with AI B2B and B2C services at 61.3 billion won, up 24.5%. Indosat Ooredoo Hutchison reports sovereign AI cloud income separately, $28 million in 2025 with guidance of $60 million in 2026, and its Zankore joint venture with Nvidia, Nokia and Ooredoo raised $3.1 billion in debt this week to build 200 MW of GPU capacity by early 2027. On the efficiency side, KPN targets around €100 million a year in operating cost savings by 2030, Proximus folds AI into a €180 million efficiency programme, and Spark credits its churn model with reducing customer loss by up to 30%. TelecomTV also cites GSMA Intelligence's finding that only 35% of telco AI initiatives include a revenue objective. Read together, the picture is that revenue is what operators now say, savings are what they can measure, and 2 of the 3 operators publishing AI revenue lines, SK Telecom and Indosat, are selling compute or cloud capacity, which is the first of the 4 credible revenue lines in the July post and the one that looks least like a telecom service. Bell has not said what its C$700 million comprises.

The number the business case is missing

Put the 2 reports side by side and the missing figure is obvious. The Index counts operators that talk about revenue and reports the revenue lines that exist. Bain estimates the cost that agentic AI adds and warns that the legacy cost may stay. Neither report, and no operator in the Index, publishes both halves for the same company: AI-attributed revenue, AI-attributed savings actually removed from the cost base, and the AI compute and token cost that Bain says will make up 20% to 30% of the operating model. Bell's C$700 million is a revenue figure with no cost of sale attached to it in public. SK Telecom's AI datacentre revenue is real, but it is a hosting business whose margin is set by power and GPU depreciation, not by the network. KPN's €100 million is a 2030 target. Bain's own preferred use cases, autonomous network incident resolution and proactive churn prevention, are the workflows where the operator has to redesign the process end to end and retire what it replaces, and they are also the workflows where the data has to be fit for an agent to act on. Deutsche Telekom's Ahmed Hafez told the AI-Native Telco Forum on 8 September that DT's network data "is not ready" for machines and that data quality demands will start appearing in supplier RFQs, which is a cost that precedes any saving. Bain's framing is the right one for the report I am completing on autonomous networks: cost per resolved incident, before and after, with the platform, compute and data cost inside the denominator. That is the number that converts a proof into a P&L entry, and it is the number no one in the Index has yet published.

What has not been published

No operator has published AI revenue, AI savings and AI operating cost for the same period. Bell has not disclosed the composition of its C$700 million or the margin on it. AT&T's 90% cost reduction is its own account, relayed by Bain, with no base figure. The Bain report itself is available from Bain; the figures here are as reported by Telecoms.com. 4 things to watch. First, whether any of the 20 revenue-first operators reports AI compute or token spend alongside AI revenue in a quarterly filing, since that would be the first complete line item. Second, whether a legacy cost is visibly retired and attributed to agentic automation, in headcount, systems or supplier spend, rather than a saving being projected. Third, the on-demand recording of the AI-Native Telco Forum session on closing the attribution gap between deployment and financial outcomes, with Deutsche Telekom, Everpure and TM Forum, which TelecomTV says will be posted within days. Fourth, whether the 2027 edition of the Index adds a financial dimension and, if it does, whether it measures cost of outcome the way Bain proposes.

Telstra Sells a Network Slice for Microsoft Teams

Telecoms.com's Nick Wood reported on 7 September that Telstra has launched Dynamic 5G for Video Calling, a business service that directs Microsoft Teams traffic onto a dedicated slice of its 5G network so that calls do not compete with other traffic during peak periods. Telstra's own description is that it "represents a move toward application-aware, business-grade connectivity, designed to prioritise performance for selected traffic, not just speeds." The operator names construction, utilities, transport and logistics as the sectors it is aimed at, where staff are on the move, temporarily located or in the field. Dynamic 5G is the umbrella brand Telstra launched for business slicing in 2025, and its distinguishing feature is a proof-of-value engine that monitors slicing traffic continuously and checks whether the customer is receiving the service promised; if it falls short, the customer does not pay. Telecoms.com sets the launch against T-Mobile US's SuperMobile business tariff, Deutsche Telekom's gaming and video-calling slices, EE's Fast Lane and VodafoneThree's SuperMobile in the UK, and cites ABI Research's forecast that the slicing market will grow from $6.1 billion in 2025 to $67.5 billion in 2030. No price, no slice parameters and no customer count were published.

3 slicing products in 3 weeks, sold 3 ways

Since 20 August, 3 operators have put differentiated network quality on sale in 3 different forms. EE sells a priority lane to consumers as a handset tariff, which I covered when Fast Lane launched. Vodafone sells a quality profile to developers through an API in Germany, without a price or a published profile. Telstra sells a slice to enterprises per application, with the application chosen by the operator and the enterprise buying the outcome. These are the 3 channels through which an operator can sell the same underlying capability, and it is useful to have all 3 on the market at once because they will produce different evidence. The tariff channel will produce an attach rate. The API channel will produce a call volume. The enterprise channel will produce a contract with a service commitment in it, and that is the only one of the 3 that forces the operator to state what the slice delivers.

The SLA clause is the news

When Vodafone launched quality on demand yesterday I wrote that the first thing a serious enterprise buyer would ask for was a service level, because a priority is not a reservation and a profile without published parameters tells the buyer nothing about what it is paying for. Telstra's answer is the proof-of-value engine. It does not publish the parameters either, at least not in what has been reported, but it commits to measuring the slice against what was promised and to not charging when the promise is missed. That is a commercially meaningful difference. It converts a best-effort improvement into a product with a defined failure condition, and it means Telstra is now generating, internally, a number no slicing launch to date has disclosed: how often a commercial slice fails to deliver what the customer bought. The pay-out rate of that engine is the honest measure of whether slicing works on a live network, and Telstra has built the instrument that produces it. Whether it will ever publish it is another matter, but the number now exists somewhere, which was not true of any of the other launches.

Application-aware means the operator picks the application

The product is sold for Teams, and only Teams. That is a practical choice, since Microsoft publishes the network endpoints its services use, which is what allows an operator to steer that traffic onto a slice without inspecting its content, and it is the application that a construction firm's site manager or a utility's field crew is most likely to be on. But it also shows where the operator's own control ends. Telstra decides which application qualifies, does the classification itself, and holds the only billing relationship. Microsoft is not a party to the arrangement, and the customer cannot extend the slice to another application without Telstra adding it. This is the same pattern that Fast Lane and the Vodafone API follow: the service is possible because it stays inside 1 administrative domain, with no counterparty to negotiate with and no shared model to agree. An enterprise that wants its own collaboration tool, its own robotics controller or its own agent to request the slice on demand, and to verify for itself that the promise was kept, is asking for the coordination layer that I argued no runtime supplies. The proof-of-value engine is a step toward it, because it is an audit trail, but it is Telstra's audit trail, read by Telstra, and the customer sees only the invoice it produces.

When Will We See Slicing as a Service?

I have been arguing for years that the natural fulfillment of the differentiated connectivity 5G promise would be slicing as a service. Specifically, network operators who will find themselves with ample capacity should create a platform that would allow third party to discover, configure, reserve and consume network resources on demand. Enterprise and government CIOs know better than telecom operators what their connectivity needs are and how they are going to evolve. They do not want to select a pre packaged slice that my correspond to some conditions or services at some point in time. They do not want to spend time educating the operator about their needs, because some of them are proprietary and become differentiating factors and they evolve over time. They are used to configuring cloud services. It is time for connectivity services to catch up.

Vodafone Launches Quality on Demand API in Germany

Vodafone announced on 7 September that its Quality on Demand network API is commercially available in Germany, the first market to get it, with more countries to follow. Light Reading's Tereza Krásová and Telecoms.com's Andrew Wooden both carried the release the same day. The API lets a business or developer select a predefined profile with specific network parameters for a customer's connection over Vodafone's 4G and 5G networks, and the release says the customer can choose the bandwidth level they need for as long as they need it. The example given is a payment provider whose card terminals keep processing transactions in a crowded stadium, concert or festival. Vodafone says it is running a proof of concept with a German broadcaster, not named, covering a football match with push-to-talk between production crew and return video over the mobile network, and it invites developers to propose applications in entertainment, transport, emergency response and remote maintenance. Johanna Wood, Vodafone's director of network APIs, said people use their phones for shopping, banking, public services and entertainment "around ten times a day" and that the API lets businesses "tailor network quality dynamically for specific use cases." The release places the launch inside the GSMA Open Gateway programme Vodafone joined in 2024, says the API follows a common industry standard so customers can scale it worldwide, and positions it alongside Vodafone's SIM Swap and Number Verify APIs, dedicated 5G slices and private networks. Telecoms.com notes that Number Verify 2.0 launched in Germany, the Netherlands and the UK in July. No price was published.

This is the launch the Open Gateway programme was built around

When EE put a consumer network slice on sale last month, I listed as 1 of 4 things to watch whether the same capability would show up as a priced quality-on-demand API through Open Gateway. 17 days later it has shown up, without a price. That is still a milestone. Quality on demand was the API the whole programme was designed to showcase, the one that would prove a network could sell a differentiated attribute to a developer rather than a bigger data bucket to a subscriber, and it has been the slowest of the 4 headline APIs to reach commercial availability anywhere. The APIs that did ship first were single-attribute lookups, number verification, SIM swap, location, the class I argued at DTW Ignite was the class that ships because it asks the network a question rather than asking it to change its behaviour. A group operator making quality on demand generally available in its largest European market moves the harder API into the same category, and that should be recorded before the rest. It is consistent with the position I set out in July, that network APIs are the earliest-stage of the credible new revenue lines, worth investing in for position while sizing near-term revenue with restraint.

The buyer is a payment provider whose customers are on 3 networks

The structural difference from EE's Fast Lane is who is paying. EE sells the attribute to its own subscriber, through a tariff, and the subscriber is on EE by definition. Vodafone sells it to a third party, and the third party's end users are spread across Vodafone, Telekom and O2 in roughly the proportions of the German market. A card-terminal vendor at a stadium can buy uninterrupted payments only for the share of its terminals, or its customers' phones, that happen to be on Vodafone. For the developer, the API is worth Vodafone's market share until somebody aggregates it, and the release addresses that by saying the API follows a common standard and so scales worldwide. A common interface is not a common network. The standard makes the developer's call portable; it does not make the other 2 German operators answer it, and the aggregation layer the operators created for exactly this purpose is not mentioned in the release. The developer proposition for quality on demand in Germany is therefore still a partial one, and the launch is best read as Vodafone establishing its own endpoint ahead of whatever the 3 operators eventually offer together.

4G and 5G tells you what the product is

Fast Lane runs on EE's 5G standalone core and moves the customer onto a dedicated slice. Vodafone's API works over 4G and 5G, which means it is not a slice. It is a policy applied to the session, a prioritised bearer with a profile attached, of the kind mobile networks have been able to set for years and have rarely exposed to anyone outside the operator. That is why it can launch nationally now rather than waiting for standalone coverage, and it is also why the release describes the result as stable and high-performing rather than as a number. A priority is not a reservation. It improves the customer's position in the queue on a congested cell; it does not guarantee what comes out of the queue, and the profile parameters that would let a buyer know what it is paying for are not published. VodafoneThree's SuperMobile slice in the UK, launched last week, attached a published minimum speed to its consumer product. The German API, sold to businesses that will build service commitments of their own on top of it, has not yet done the same, and that is the first thing a serious enterprise buyer will ask for.

It ships because it stays inside 1 estate

The pattern from Fast Lane holds here in a different form. Quality on demand as launched is a bounded request inside a single administrative domain: 1 operator, 1 profile, 1 session, 1 billing relationship with the developer. No counterparty negotiates the profile, no ontology has to be agreed across an ownership boundary, and no audit trail has to be accepted by 2 parties. That is the class of capability I argued no runtime supplies the coordination for and that therefore ships only when the coordination is not needed. The question the API opens, and does not answer, is what happens when the buyer is not a developer but a system: a checkout agent, a delivery routing engine, a broadcaster's production controller deciding per session whether a given profile is worth its price against a given cell load. At that point the network's AI and the customer's AI are negotiating a resource across a boundary, and the API gives them a verb without giving them a shared model of what the verb commits either side to. Vodafone's release does not address that case, and it is the case its enterprise customers will bring first.

What has not been published

No price, no pricing unit, no profile parameters, no service level or remedy, no named paying customer, no statement of what the profile delivers on a cell that is already saturated, which is the only condition under which the product matters, and no volume figure from the July Number Verify launch that would indicate what developer uptake of Vodafone's APIs looks like. The broadcaster proof of concept is unnamed and uncosted. 4 things to watch. First, a price list, because the moment quality on demand has a published unit price it becomes possible to compare the developer channel against the tariff channel EE chose. Second, whether the 3 German operators offer a single quality on demand call, through the aggregator or bilaterally, since that is the point at which the payment provider's business case stops being a fraction. Third, the first named enterprise customer with a transaction volume, which would be the first demand-side evidence for the API the programme was built around. Fourth, whether Vodafone Germany also sells the same profile to consumers as a tariff, because an operator that does both is telling the market which channel it thinks the capability belongs in.

TelecomTV Publishes First AI-Native Telco Index

TelecomTV published its first AI-Native Telco Index on 7 September, the day before its AI-Native Telco Forum opens in Düsseldorf. Ray Le Maistre's article describes a 276-page report assessing 56 operators against 4 dimensions, foundation readiness, strategic intent, execution evidence and transformation velocity, each made up of 5 indicators, using published evidence only, with a cut-off of 30 June 2026. The index assigns each operator 1 of 7 archetypes rather than a rank; the decision not to publish a league table was taken with the ANTA steering board, whose members come from Axiata, Deutsche Telekom, Orange, NTT Docomo and Rakuten Mobile. 9 operators are placed in the top archetype, AI Vanguard: AT&T, China Mobile, Deutsche Telekom, KDDI, NTT and NTT Docomo, Rakuten Mobile, SoftBank Corp, Telefónica and Verizon. 5 are Infrastructure Architects, 15 are Strategic Accelerators and 1 is at Transformation Pending. The stated rule is that "announce does not mean executed": a commitment is recorded as a commitment, and a deployment is recorded only where the operator has published something showing it is running. SK Telecom, which the article notes is often named the leading AI-native operator, is a Strategic Accelerator because many of its efforts were still at the planning stage at the end of June. A 12-page executive summary is free; the full index is available to ANTA partners and licence holders. This post works from TelecomTV's article; neither the summary nor the full report was read for it.

Most assessments of operator AI progress, including most vendor-sponsored surveys, score intent: budgets planned, priorities ranked, use cases identified. This index scores what has been published as running, and it records an announcement as an announcement. That is the distinction between field-validated and commercially validated that AI-RAN claims have needed for 2 years, applied here across the whole operator estate, and the SK Telecom result shows the rule working. An operator with a clear strategy, disclosed investment, revenue already generated and a large revenue target still lands below the top tier because its published evidence at the cut-off described plans rather than operations. Producing that result about the operator most often cited as the leader, in a report whose steering board includes 5 operators, is a sign the methodology was allowed to run.

No financial dimension yet

The 4 dimensions measure readiness, intent, execution and speed. None measures return. An operator reaches AI Vanguard by publishing the most evidence that AI systems are in production across its operations, and that is a real achievement, but it is the same class of evidence as the customer-experience award that TM Forum gave Google Fiber last week: proof that it works, not proof of what it is worth. The index is explicit that it measures transformation progress and not who is better, and that framing is useful, because it means nobody should read the Vanguard list as the operators making money from AI. I argued in July that most of what operators call AI monetization is cost avoidance, and that the discipline starts with refusing to aggregate cost saving, defended revenue and new revenue; an index that counts deployments without asking which of the 3 flows each one serves cannot make that separation either. The industry now has a reliable way to count deployments. It does not yet have a published operating cost per subscriber, per activation or per ticket, before and after, from any of the 9, and until one of them publishes that figure the index measures the input side of a business case whose output side remains an estimate.

Because the method admits only what an operator has published, an operator that publishes more scores higher, other things being equal. 4 of the 9 Vanguard operators are Japanese, 2 are American, 1 is Chinese and 2 are European, and 2 of the 9, NTT Docomo and Rakuten Mobile, sit on the steering board that shaped the method. The article says archetype assignment is an editorial judgement based on the evidence analysed, and there is no reason to doubt that. The point is narrower: operators with a corporate habit of publishing technical detail, and operators for whom the network is a marketing asset, will be over-represented at the top of any evidence-based index, and operators that run production AI quietly will be under-represented. The index cannot correct for that without abandoning its rule, and it should not abandon its rule. The caveat applies to every evidence-based index, this one included.

Everything the index can see is inside 1 operator's estate: its data foundations, its stated strategy, its deployments, its pace. That is the version of autonomy that ships, and it is where the evidence should be collected first. The next stage, in which an operator's network AI has to coordinate with an enterprise AI, a cloud provider's AI or another operator's AI, produces almost no published evidence yet, because the coordination layer that would make it work is the one I argued no runtime supplies and that NGMN has since itemised even for coordination inside a single operator. An index built on published deployments will show that stage as empty for some time, and the absence will be accurate. A future edition could usefully add an indicator for it: whether the operator has published any interface, ontology or audit model through which an external agent can act on its network, since the developer APIs alone do not answer that question.

What has not been published

The per-operator scores, the weighting of the 20 indicators, the definition of "running" used for execution evidence, and whether any indicator captures financial disclosure are all in the licensed report and not in the article. The 56-operator list is published; the archetype of each operator outside the 3 named tiers is not. 3 things to watch. First, whether the 2027 edition adds a return dimension, or an indicator for published cost or revenue attributable to AI, which would be the single most useful change to the method. Second, whether any of the 9 Vanguard operators publishes an operating cost figure for an automated domain against its pre-automation baseline in the next 12 months, since they are now on record as the operators with the most to show. Third, where SK Telecom lands next year: the article expects it to be among the strongest profiles, and if the planned efforts have become published deployments by then, the index will have shown it can measure movement and not only position.

Ericsson Says the Telco Edge Was Too Early

Ericsson has put a name to the failure of the last edge computing cycle. Joe Constantine, the company's Americas chief strategy and technology officer, told Fierce Network in a piece published by Diana Goovaerts on 21 August that mobile edge computing was early rather than wrong, and his summary of what went wrong is short. "Ten years ago, MEC was a supply side concept." What has changed, in his account, is that a demand driver has arrived. "We have AI inferencing. It's the growth opportunity, an application that MEC did not have at the time. So, if you look at this, we believe that the industry, the technology and the market is vastly different today from 10 years ago." He supports the traffic case with Ericsson's own forecasts, that global mobile traffic will triple between 2023 and 2029 with AI as the key driver, and that uplink traffic will grow 10 times by 2035. He argues the network has changed too, from best effort connectivity to 5G built for time-critical communication and capable of 15 millisecond latency and "five nines of reliability." Fierce notes in the same piece that TM Forum's chief executive has told it that operators should not bet the farm on the edge.

The diagnosis is correct and it is unusual to hear it from a vendor

Supply side concept is the right post-mortem, and it is the shortest accurate description of that decade anyone has offered on the record. Operators built edge capacity because they had property, power and a latency story, then went looking for someone who wanted it. The platforms worked. The buyers did not appear. When I wrote about a fully programmable multi-access edge platform in November 2018, the services I could name were faster file uploads, console games without a console, and editing a document without downloading it. Those were real improvements to existing experiences. None of them was a business that an enterprise procurement department was going to sign for. Inference is a materially better answer than that, because it is a workload with a measurable cost that somebody is already paying somewhere else. That is a genuine change in the argument and it should be credited before anything else is said about it.

15 milliseconds is a central office number

The interesting discipline in Constantine's case is that his own numbers settle the location question, and they settle it away from the radio. A 15 millisecond budget is a loose one. It is comfortably met from a metro exchange or a mobile switching office serving hundreds of sites, and it does not require compute at the tower. That matters because the edge conversation still routinely conflates two very different capital programmes. Putting accelerators in a few hundred central offices is a brownfield project on estate that is already zoned, powered, cooled, fibred and physically secured. Putting them at tens of thousands of cell sites is a different business with a different power bill, a different maintenance model and a different landlord. I have argued on the AI Grid that deployment starts at central offices and mobile switching offices for exactly these reasons, and the 2018 platform was built into the central office for the same ones. Nothing in the Ericsson case contradicts that. If 15 milliseconds is the requirement, the requirement is an exchange.

The economic claim is asserted rather than costed

"Routing all this traffic to a centralized cloud isn't just only slow, it's economically not even sustainable" is the load-bearing sentence of the whole argument, and it arrives without a number attached. There is no published cost per inference at a telco edge site set against the same inference in a hyperscale region, at comparable utilisation, including the operator's power, cooling, refresh and remote-hands costs. Until that comparison exists, the economic case for distribution rests on the intuition that moving bits is expensive and moving them less must therefore be cheaper. That intuition ignores utilisation, which is what actually decides the economics of accelerators. A central region runs its fleet hot across many customers and time zones. A metro edge site runs a smaller fleet against local demand that peaks and troughs. The traffic forecasts are also Ericsson's own, published in its Mobility Report, and a 10 times uplink projection reaching to 2035 is a forecast rather than an observation. It should be read as one.

Physical AI names a use case, not a counterparty

Asked what actually requires the edge, Constantine points to robotics and physical AI. "Any system that moves, so drones, vehicles, humanoids, robots talking to other robots, all these will become an autonomous system that needs a network to think with it." His example is an autonomous car whose onboard sensors can see what is in front of it but not around the corner, with network sensing supplying the rest. The technical case is sound. The commercial case has the same shape as the one that failed. A drone fleet, a warehouse robot estate and a vehicle platform are owned by companies that are not the operator, and the thing being sold to them is not raw compute, which they can buy cheaper elsewhere, but contextual awareness that the network holds and they do not.

Selling that is a coordination problem before it is a latency problem. Robots talking to other robots across an ownership boundary need an agreed model of what is being asserted, whose authority stands behind it, and what record both sides will accept afterwards. That is the gap I argued no runtime supplies, and that NGMN has now enumerated at length for coordination inside a single operator's own domains, which is the easier version of the same problem. An operator can install the accelerators in its exchanges this year. It cannot unilaterally produce the model that lets a vehicle manufacturer's autonomy stack trust and pay for what the network says about the road. The compute is the part of this that money can buy quickly, which is why it is the part vendors describe in detail.

What to watch

3 things. First, where the accelerators physically land. If operator edge AI deployments in the next 18 months are announced at central offices and metro sites, the location argument is settled and the cell site edge can be retired from the conversation. Second, a published cost per inference or cost per token at a telco edge site against a hyperscale baseline, from anybody, on the record. That single figure decides whether distributed inference is an economic position or a latency preference. Third, the first contract in which an enterprise pays an operator for network context rather than for compute or connectivity, because that is the class of deal the 2010s never produced and it is the only one that would show the demand side has actually changed.

EE Launches First Priority Consumer Network Slice

EE launched a service called Fast Lane on 20 August, and Telecoms.com reported it as the UK's first commercial network slice sold to consumers and small businesses. The mechanism is simple. When congestion is expected, the customer is moved onto a dedicated slice of EE's 5G standalone network, which the operator brands 5G+. The scenarios EE names are rush hour, major events, live streaming from a sold-out gig, taking card payments at a festival, joining a video call while travelling. Claire Gillies, chief executive of BT's Consumer Division, called it "the UK's first commercial network slice to deliver meaningful benefits to both consumers and businesses." EE says 5G+ now reaches around 54 million people, 78% of the UK population, with a plan to reach 99% by the end of March 2030, as part of a £40 billion investment programme. Fast Lane is sold only through a premium handset plan, not as a standalone or SIM-only option.

I want to start by emphasizing the innovative aspect. This is a real new revenue line. It is not a framework paper, it is not a pilot inside one operator's walls described as a capability across everyone's. Somebody is going to pay money to EE for a network attribute, and after 8 years of the slicing conversation producing conference sessions instead of invoices, that deserves to be said plainly. It is also the first time I can recall a British operator putting a retail price on differentiated network quality rather than on a bigger bucket of gigabytes. That is a category change and it is worth taking seriously.

The product is priced on scarcity, and the scarcity is the operator's own congestion

Now the rationale. Fast Lane sells preferential treatment at a congested cell. The value of the product is a direct function of how congested that cell is. Telecoms.com raised the obvious question, whether consumer slicing gives operators a reason to slow densification, and then answered it fairly by pointing at EE's own build plan, which is running hard from 78% coverage toward 99% and is backed by a £40 billion commitment. I do not think EE is withholding capacity. But the accounting point survives the good intentions. Willingness to pay for Fast Lane is highest exactly where the network performs worst, and it falls as the build succeeds. That is not a scaling revenue line. It is a harvest on a constraint, and the constraint is one the operator has publicly promised to remove.

A congestion-priced lane is a legitimate thing to sell and a bad thing to extrapolate. If I were modelling this, I would treat the addressable moments as a fixed and slowly shrinking pool: the stadium, the festival, the commuter peak, the trade show. Those are real, they recur, and they do not compound. The number to watch is not the launch, it is whether the attach rate holds in year 3 in the places where 5G+ has since been densified. That number is the only one that would tell you whether this is a product or a symptom.

Quality on demand shipped as a tariff because a tariff needs no shared model

Here is the part I find genuinely instructive, and it lands on an argument I have been making for a year. Quality on demand is 1 of the 4 showcase network APIs the GSMA has put at the centre of Open Gateway, alongside device location verification, SIM swap detection and number authentication. The industry has spent 3 years and roughly 300 mobile networks trying to sell that capability to developers through an interface, and this month the answer to why it has not converted arrived in the form of a systems integrator joining the programme. Meanwhile EE is selling the same underlying capability to a consumer, through a handset tariff, and it works.

The reason is the one I set out when I argued at DTW Ignite that the API is not enough. Fast Lane requires no shared model with anybody. One operator, one subscriber, one radio, one contract, one billing relationship. There is no counterparty to negotiate with, no ontology to agree, no authority delegated across a boundary the operator does not own, and no audit trail that two parties both have to accept. It is a unilateral act inside a single administrative domain, which is precisely the class of capability that ships. The Open Gateway APIs that shipped are all in that same class, single-attribute lookups and bounded requests inside one operator's estate, and I have argued that is part of why they are worth what they are worth. EE has just proved the rule from the other direction. Take the exact same network function, strip out the requirement to coordinate with an external party, and it goes to market in a quarter rather than a decade.

That could be read as a warning. The capabilities that clear the boundary problem are the ones an operator can sell to its own subscriber. Everything on the other side of the boundary, the enterprise AI negotiating a service level with a network AI, the slice provisioned and assured across two estates, still needs the meta-model of topology, ontology, authority, state and audit that I argued no runtime supplies and that NGMN has now enumerated at length. Fast Lane does not advance that work. It circumvents it.

Selling it only with a handset tells you what EE thinks it has

The packaging is the tell, and it is the detail I would push hardest on. Fast Lane is available only on a premium handset plan. Telecoms.com flagged that this may hamper uptake, which is true, but the more interesting reading is what it says about internal conviction. If you believe you have built a network capability, you sell it as an attribute of the connection, on any SIM, at a price, and you let the market tell you what it is worth. If you believe you have built a retention lever, you bolt it onto the highest-value handset tier where it defends an existing margin and never has to survive a standalone price test. EE has chosen the second. That is a rational commercial decision and it is also an admission that the capability is not yet trusted to stand on its own.

What to watch

4 things. First, whether EE ever offers Fast Lane SIM-only. The day it does, the company believes it has a network product; until then it has a device upsell with a network feature attached. Second, whether any operator anywhere publishes an attach rate or an ARPU delta for consumer slicing, because the launch is easy and the second year is the evidence. Third, whether the same capability shows up as a priced quality-on-demand API through Open Gateway at a comparable value, which would be the first real read on whether the boundary tax is worth what I think it is worth. Fourth, whether a lane sold on congestion sits comfortably inside the UK's net neutrality framework once it has scale rather than novelty. I am not going to declare a verdict on that one, and I would be surprised if nobody asks the question.

The industry has spent 8 years promising that slicing would let operators sell different connectivity products rather than by the gigabyte. The first commercial consumer version in Britain sells one attribute, at specific times, bundled with a phone. That is progress, and it is a fraction of the size of the story that was told to justify 5G standalone. 

NGMN Confirms Agentic AI Needs Improvements

The Next Generation Mobile Networks Alliance published a report on 12 August, "Network Automation and Autonomy Phase III: Agentic AI for Autonomous Mobile Networks," and it reads like a standards body catching up to an argument I have been making for a year. The headline finding, reported by Keith Dyer at The Mobile Network, is that the industry has to solve interoperability, security, governance and data before agentic AI can deliver autonomous networks at commercial scale. The detail underneath the headline is what matters. NGMN says the current ecosystem is too fragmented for agents to form a consistent understanding of the network, because vendors use different data models and different terminology, and it calls for common information models, ontologies and semantic mappings, aligned across 3GPP, TM Forum, ETSI, O-RAN, IETF, W3C and BBF. It asks for a telecom-grade Zero-Trust Agent Ecosystem with secure agent identity, authentication, authorisation, policy enforcement, audit logging, runtime monitoring, and the ability to revoke or quarantine an agent. Read that list again. Topology, ontology, authority boundaries, state, audit trails. That is the meta-model of the agentic plane, and it is now in a report with an operator alliance's name on the cover.

I want to be precise about the provenance here, because the sequence is the point. I argued at DTW Ignite that the API is not enough, that the interfaces we built for developer access were never designed to let an autonomous agent understand what it is acting on. I argued a few weeks later that the agent runtime is not the agent model, that a place to execute an agent and put guardrails around it supplies none of the topology, ontology, authority and audit that coordination between agents actually needs. NGMN has now enumerated eight cross-organisation challenge areas, and they map almost one for one onto that case. Lack of standards for agent knowledge and context sharing. Incomplete information modelling for networks. Lack of end-to-end security, identity and trust for autonomous functions. Lack of operator-friendly governance and lifecycle tools. Under-addressed economic and organisational readiness. When a body chaired by Orange's group CTO writes the same list I published, the argument stops being a contrarian read and becomes the consensus. That is a good day for the thesis. It is a better day to point out the part the report does not solve.

The report describes coordination inside a boundary the operator owns

Look closely at the architecture NGMN proposes. A supervisory agent coordinates specialist agents, one each for RAN, core, transport, cloud and security, to diagnose a service problem, weigh corrective actions, resolve conflicts, and verify that service was restored. The worked examples are fault management, service assurance and RAN optimisation. In the RAN case, a cross-domain agent delegates intent to a RAN optimisation agent, which supervises a sub-agent in the infrastructure layer, and when an optimisation action collides with network-wide energy management the cross-domain agent arbitrates the trade-off. This is a serious and correct piece of engineering. It is also, in every example, coordination across domains that a single operator owns. RAN, core, transport, cloud and security are five administrative domains inside one company's walls. The hard thing NGMN is describing is getting an operator's own silos, built by different vendors with different data models, to hand each other a shared and trustworthy understanding of state. That is worth doing, and it is exactly why a runtime and a set of guardrails were never going to be enough. But it is one owner reconciling with itself.

The problem I have been pointing at sits one boundary further out, and NGMN's own scope is the clearest evidence that it is a distinct and harder problem. When an enterprise's AI wants to negotiate with an operator's network AI, for a service level, a slice, a capacity commitment, a remediation, there is no shared owner to impose a common ontology, no single authority to issue the identities, and no single audit log that both parties will accept as the record of what was agreed and who was answerable. Everything NGMN specifies, the semantic alignment, the zero-trust identities, the audit trail, is defined for agents operating within the operator's estate. Across an ownership boundary, the same requirements do not disappear. They get harder, because now two parties have to agree on the model before either can trust an action, and neither controls the other. NGMN has validated that the meta-model is necessary. It has not, and by its own framing could not, close the gap where the two hardest words in this whole subject live, which are coordination and accountability between parties who do not report to the same CTO.

Containment, coordination, accountability, in that order of difficulty

It is worth keeping the three layers separate, because vendors and now standards bodies keep collapsing them. Containment is the Zero-Trust Agent Ecosystem: the identity, the authorisation, the runtime monitoring, the kill switch that lets you revoke or quarantine an agent inside your own network. That is genuinely useful and NGMN is right to specify it in detail. Coordination is the shared ontology and semantic mapping that lets one agent understand what another agent is doing, and this is the layer NGMN is now trying to build across an operator's internal domains. Accountability is the ability to produce, on demand, the identity of an agent, the authority under which it acted, and a trail that a regulator and an outside counterparty would both sign, across a boundary you do not own. The industry has spent a year shipping the first layer and calling it the third. NGMN has now, to its credit, put real weight on the second. The third is still open, and it is the one I argued the EU AI Act made expensive when its enforcement powers switched on at the start of this month, because enforcement does not stop at the edge of your administrative domain.

There is a discipline point here that operators should not lose in the enthusiasm of seeing a favourite thesis ratified. NGMN itself flags it: "While technology groups focus on architecture, few have formal guidance on the required telco organisational changes," and it lists economic and organisational readiness as an under-addressed challenge in its own right. The meta-model is not a product you procure. It is a set of agreements about topology, ontology, authority and audit that has to be built and then adopted, and the report's call for alignment across seven standards organisations is a fair description of how long that takes. Cross-SDO harmonisation of information models is measured in years, not quarters. When I helped set network autonomy targets in operator programmes, the technology was rarely the thing that slipped. The organisational change and the cross-domain agreements were. Anyone booking autonomous-network savings against a framework paper rather than a ratified model is doing the same thing I warn against when vendors stack their individual efficiency claims into a total no network has ever achieved. Read this report as the industry agreeing on the destination. Do not read it as arrival.

What to watch

The honest test has not changed, and NGMN has now made it a formal requirement rather than a personal opinion. For any action an agent takes on its own, can you produce its identity, the authority under which it acted, and an audit trail that both a regulator and an outside counterparty would accept. If the answer lives entirely inside one vendor's runtime, you have containment and you should not mistake it for accountability. NGMN has described the plane the runtime runs on, in more detail than anyone in the industry has committed to print, and it has correctly located the work in standards and organisation rather than in any single box. The next phase belongs to whoever builds the ontology and the identity fabric that hold up not just across an operator's own domains, but across the boundary to the enterprise on the other side of the negotiation. That boundary is where the value is, and it is the one the report stops at. Watch for whether the alignment NGMN asks for actually happens across those seven bodies, and watch, as always, for a pilot inside one operator's walls being described as autonomy across everyone's.

Agentic AI: Autonomous Agents Governance and Enforcement

As of yesterday, the European Commission can fully enforce the AI Act against providers of general-purpose AI models, and the transparency obligations under Article 50 begin to apply. The obligations themselves have been in force for a year. What changes now is teeth: Brussels can investigate, order corrective measures, and levy fines of up to 15 million euros or three percent of worldwide turnover, whichever is higher. Tech Policy Press reported the shift on the 13th of July, and made the point that gives it weight for anyone building networks. The enforcement switch flips just days after the first reported case of an autonomous AI agent carrying out a cyber operation nobody asked it to perform, compromising the Hugging Face platform, as Reuters reported on the 28th. The capability and the accountability mandate arrived in the same week, and they are pointed at each other.

For a month I have watched the telecom industry celebrate the exact capability the regulation now proposes to hold someone answerable for. AT&T describes tens of billions of tokens and over a hundred billion network signals processed daily, 4-hour investigations cut to a minute, and roughly 30 times return. TM Forum reports the sector crossing into Level 4 autonomous networks. Vendors are shipping the agentic NOC and the self-healing network. The good news and the governance problem are the same sentence. Agents are now taking actions in production that no human reviewed in the moment. That is the point of them. It is also the point at which a regulator asks who is answerable.

Enforcement is not an abstraction. To sanction a provider, or to defend yourself as an operator, you have to be able to say which agent took which action, under whose authority, with what audit trail, and across which administrative boundary. In a telecom network that last clause is the hard one. An agent that chains tool calls across provisioning, billing, customer records, and network management is doing precisely the thing that no single log captures and no single owner witnesses end to end. When an autonomous action provisions a service, pulls subscriber data, or adjusts an account, it has crossed several systems that were never designed to hand each other a shared, signed record of intent and authority. The action happened. Reconstructing who owned it is the part that does not exist yet.

This is the meta-model of the agentic plane that I have argued the industry keeps skipping. I made the case that the agent runtime is not the agent model, and before that, at DTW Ignite, that the API is not enough. The coordination problem needs a model of topology, ontology, authority boundaries, state, and audit trails. A runtime supplies none of those across a boundary. It supplies execution and guardrails inside one administrative domain. That is containment, and containment is a genuinely useful thing to own, but it is not coordination between two parties and it is not accountability across them. The regulation just made the distinction expensive.

Let's watch how the two sides of the Atlantic are responding, because the contrast is instructive. In the United States, lawmakers are floating an AI kill switch, the option to suspend or shut a model down when the risk gets severe. A kill switch answers one question well: can I stop it. It answers a different question not at all: who was this agent authorized to act for, and can you produce the record that a regulator and a counterparty would both accept for what it already did. Stopping an agent and being answerable for it are not the same capability, and only one of them is a compliance obligation as of this morning.

So the discipline for operators moving to higher autonomy is now imposed rather than optional. Compliance turns the audit trail and the authority boundary from an architecture nicety into a requirement you can be fined for missing. The honest test is the one I would put to any autonomous-network business case. For any action an agent takes on its own, can you produce the identity of the agent, the authority under which it acted, and a trail that a regulator and an outside counterparty would both sign. If that record lives entirely inside one vendor's runtime, you have containment, and you should not mistake it for accountability. Deloitte's 2026 enterprise survey found only about 1 organization in 5 reports a mature model for governing autonomous agents. The regulator did not wait for the other four.

The agents learned to act this summer. Today the law asked who authorized it. The operators who will be interesting in this next phase are not the ones with the most agents in production. They are the ones who can answer that question about any one of them, on demand, across a boundary they do not own. That answer is not a feature of a runtime. It is the plane the runtime runs on, and the industry has been shipping the second while promising the first.