How the AI Act proposal tied risk to the use, whatever the model

The week

On April 21, 2021, the European Commission published its proposal for a regulation laying down harmonised rules on artificial intelligence, the AI Act. It took a risk-based approach. A short list of practices would be prohibited, such as social scoring by public authorities. A defined list of high-risk uses would face requirements on risk management, data governance, technical documentation, record-keeping, transparency, human oversight, accuracy and robustness. That list included AI used to evaluate the creditworthiness of individuals, to decide access to public benefits, in recruitment and worker management, and in certain law enforcement settings, as well as AI in products such as medical devices that already needed third-party conformity assessment. Some systems, such as chatbots, would face transparency duties. Most other AI would face no new obligations.

Most early commentary on the proposal focused on its definition of AI, which was broad enough to cover machine learning, logic-based and statistical approaches. That breadth mattered less than it seemed, because the definition alone triggered nothing. Obligations attached to the use. The same gradient-boosted model could be unregulated in a marketing team and high-risk in a lending team. For regulated organizations, that design had a practical consequence: classification had to be done process by process, by people who understand the processes. The proposal’s four tiers give the structure: prohibited practices, high-risk uses, transparency duties, and everything else.

Prohibited practices: narrow, but aimed at the public sector

The prohibited list was short. It covered AI that manipulates people through subliminal techniques, AI that exploits the vulnerabilities of specific groups to cause harm, social scoring by public authorities, and real-time remote biometric identification in publicly accessible spaces for law enforcement, subject to exceptions.

For banks and hospitals, this tier was mostly a check, not a program. For public agencies operating in or selling into the EU, it was more pointed. Social scoring by public authorities covered systems that rate people on social behavior or personal characteristics and then treat them unfavorably in unrelated contexts. A benefits fraud model that draws on data far removed from the benefit in question sat uncomfortably close to that line. Agencies needed a documented view on where their risk-scoring systems stood.

High-risk uses: where the work concentrated

The high-risk tier carried the heavy requirements, and it was defined by function. In financial services, creditworthiness assessment of individuals was listed. In healthcare, AI in medical devices fell in through existing product safety law, which tied AI obligations to the device approval process. In government, AI that decides eligibility for public assistance, prioritizes emergency dispatch, or supports certain law enforcement decisions was listed. Recruitment and worker management were listed for every sector, which meant a hospital or bank using AI to screen applicants was in scope even if it used no AI with customers or patients.

The requirements fell mostly on whoever developed the system: a risk management process, training data that is relevant and representative, automatic logging, technical documentation detailed enough for an authority to assess, human oversight designed into the system, and declared levels of accuracy and robustness. Users of high-risk systems had narrower duties, such as following instructions for use, monitoring operation and keeping logs.

Because the tier followed function, the hard classification questions were about process boundaries. Is a model that pre-fills a loan officer’s recommendation a creditworthiness assessment? Or is it decision support outside the list? Is a triage tool in a hospital a medical device, or administrative software? These are not questions a data science team can answer alone. They need the credit policy owner, the clinical lead and counsel.

The stakes were set high enough to force those conversations. The proposal allowed fines of up to 6 percent of global annual turnover for the most serious breaches, and its reach followed the affected person, not the server. A US lender scoring EU residents, or a life sciences company screening applicants for European sites, could be in scope without any EU infrastructure.

Transparency duties: telling people they are dealing with AI

The third tier imposed disclosure without the full high-risk regime. People had to be told when they were interacting with an AI system such as a chatbot, unless it was obvious, and when emotion recognition or biometric categorization was applied to them. Synthetic images, audio or video depicting real people or events had to be labeled.

For customer service and citizen service teams, this was the tier most likely to touch daily operations. A bank’s virtual assistant, a health system’s symptom-checker chat and an agency’s benefits helpline bot would all need clear disclosure. The cost was low. The design question was how to disclose without degrading the service, and how to offer a route to a person.

Everything else: unregulated, not unimportant

Most AI in use fell into the fourth tier and faced no new obligations. Fraud detection, document classification, IT operations analytics and marketing models generally sat here, though the proposal encouraged voluntary codes of conduct.

Two cautions applied. First, systems move between tiers when their use changes. A document classification model reused to route benefits applications can drift into a listed use without any change to its code. Second, other law still applied. Data protection rules, sector supervision and anti-discrimination law did not depend on the AI Act’s tiers. An organization that treated the fourth tier as a safe harbor misread the proposal.

The use-based design suggests an operating model. Each business process owner registers the AI used in their process and states its purpose, and a central team classifies uses and tracks any change of purpose. That puts accountability where the knowledge sits, and it survives changes to the legal text, because every later version kept the link between use and obligation.

With hindsight

From 2026, the use-based core survived the legislative process largely intact while much else changed. The final AI Act was adopted in 2024 and entered into force on August 1 of that year. Generative AI prompted obligations for general-purpose AI models that were not in the 2021 text. “User” became “deployer”, with added duties such as fundamental rights impact assessments for public bodies and certain deployers in credit and insurance. The final high-risk list added life and health insurance pricing and carved fraud detection out of the creditworthiness category. Prohibitions applied from February 2025 and general-purpose AI obligations from August 2025. In 2026, after the Commission’s late-2025 omnibus proposal, EU lawmakers agreed to push the main high-risk deadlines back to late 2027 and 2028; organizations should confirm the current dates with counsel. Organizations that classified by use early are managing a schedule change, not starting from zero.