Citcom AI Trustworthiness Label
The Citcom AI Trustworthiness Label is a Citcom.ai service, currently under pilot, through which TEF site partners independently assess AI systems deployed in smart-city contexts — across mobility, energy, citizen services, urban planning, and related domains.
Its purpose is to provide a trusted, recognisable signal: an independent expert opinion on whether an AI system has been developed and operated responsibly for its intended use. The Label gives cities a sounder basis for procurement and gives providers a credible way to demonstrate responsible practice on the market.
Not a conformity certificate. The Label and its recommendations do not constitute a conformity certificate and are not legally binding — in particular, they are not a conformity assessment under the AI Act. The Label is an expert, evidence-based judgement, not a pass/fail certification. Its value lies in the independent third-party assessment itself, which builds trust between cities and providers and eases the procurement process.
Who requests an assessment? Two streams
An assessment can be initiated from either side of the market. Both lead to the same independent process and the same kind of report.
Stream 1 — Municipality. A municipality needs to procure an AI solution but lacks the in-house capacity or competence to evaluate it. It requests an independent third-party assessment from a TEF site to support its procurement decision.
Stream 2 — AI provider. An AI provider requests the assessment voluntarily, so that its solution becomes more credible on the market and easier to recommend to public-sector buyers.
How it works: the value proposition
A TEF site conducts the assessment as an independent domain expert and releases an assessment report. The TEF site is not auditing compliance against a standard — it is forming a professional, evidence-grounded opinion about whether the system has been developed and operated responsibly in its smart-city context.
The tangible outcome of the process is a system of badges. Each badge corresponds to a specific dimension of trustworthiness that was assessed. Badges are awarded independently — a provider may receive one, two, or all three — and receiving a badge on one dimension says nothing about the others.
| Badge | What it covers |
|---|---|
| Technical Testing | How the system was evaluated for trustworthiness: testing methodology, independence of testing, performance, fairness, robustness and failure modes, safety, and explainability. |
| Governance | The governance, oversight, and operational practices around the system: risk management, human oversight, accountability, monitoring and incident response, and documentation. |
| Impact | The effects of the system on people, communities, and society: affected populations, differential impact, transparency and recourse, broader societal effects, participation, and environmental footprint. |
Each badge would include a watermark to ensure authenticity and prevent misuse, and would be verifiable through the Citcom Hub, allowing external stakeholders to confirm its origin and evaluation status.
Customised to context and use case
Each evaluation can focus on specific aspects and is customised to the deployment context and the actual use case of the AI system. The questions addressed for each badge represent a minimal set; a TEF site may add further ones where the specifics of the system or its smart-city context warrant it. This is what makes the assessment meaningful rather than a generic checklist.
Consistency across TEF sites
Because each evaluation is customised — and because different TEF sites contribute different specialised expertise — it is essential that all sites follow the same guidelines. Shared guidelines are what keep the Label consistent and equally valuable wherever it is awarded, so that a badge means the same thing regardless of which TEF site issued it.
The two reference documents below define and illustrate that common approach:
- Assessment Guidelines for TEF sites — how to conduct an assessment, what each badge covers, and how to exercise and document expert judgement consistently.
- Example Assessment Report — a fully worked (fictional) report showing how the guidelines are applied in practice and how a finished report looks.
For the Technical Testing badge, one such resource is the AI Assessment Sandbox Configurator, developed under the Luxembourg AI Factory. It offers a harmonised way of configuring and conducting the technical evaluation, and is suggested as one option among others — TEF sites remain free to apply their own methodologies and tools where these better fit the system and context.
Multiple TEF sites for a single solution
If a solution would benefit from complementary expertise available across several TEF sites, an AI provider can undergo assessments in multiple locations. In such cases, the first-contact TEF site coordinates the overall process: it connects with the additional sites (which assess independently), ensures each manages its own contractual and operational responsibilities, consolidates the results into a unified report, and oversees the issuance of the corresponding badges.
Assessment Catalogue
The specialised methodologies, tools, and test suites available across the network can be consulted in the AI Assessment Catalogue:
The catalogue helps innovators understand which capabilities are applied to their systems, and helps cities see how specific trust dimensions are assessed.