High Risk under the AI Act: What It Means for Manufacturing

High-Risk Classification under the AI Act: What It Means for Manufacturing

The AI Act (the EU Regulation on Artificial Intelligence) introduces a risk-based approach that imposes stricter obligations on AI systems classified as "high risk". This classification is not automatic for every use of AI, but applies to systems that pose a significant risk of harm to people's health, safety or fundamental rights.

In the context of manufacturing companies and critical industrial sectors, the "high risk" classification is particularly relevant and is triggered mainly in two scenarios defined in Annex III of the Regulation:

1. AI Systems for the Safety and Reliability of Critical Components

AI systems that act as safety components in products subject to European Union harmonisation legislation are considered high risk, such as:

Machinery and Robotics: AI systems used for safety functions (for example, speed control, emergency stop, collision prevention) in machines, lifts, pressure equipment and toys. If the AI contributes to the safe operation of these products, it falls within the high-risk category.
Safety Devices in Critical Environments: AI systems embedded in protective devices or complex machinery that, in the event of a malfunction, could cause serious injury or death.Example: An AI system that autonomously manages the emergency braking of a forklift or that controls the pressure in a chemical reactor.

2. AI Systems for Human Resources and Workforce Management

Another critical area in manufacturing is the use of AI in workplace contexts that affect employees' rights. Systems used for the following purposes fall within the high-risk category:

Recruitment and Selection: AI systems designed to analyse and filter CVs or to assess applications.
Performance Monitoring and Evaluation: Systems that make decisions, or significantly support decisions, regarding promotion, termination of employment, task allocation or the monitoring and evaluation of workers' performance.Risk: In these cases, the risk relates not only to physical safety, but also to the right to non-discrimination, to dignity and to fair working conditions.

Consequences of the "High Risk" Classification

When an AI system is classified as high risk, the company (in this case the provider that develops the system or the user/deployer that adopts it) must comply with a series of strict obligations before placing it on the market or putting it into service.

Area of Obligation	Detailed Description
Risk Management	Implementation of a risk management system to identify, analyse and mitigate the system's risks throughout its entire life cycle.
Data Quality	Obligations on the quality of training, validation and test datasets, to minimise bias and ensure accuracy and robustness.
Transparency	Providing clear and detailed instructions for use to end users, with information on the system's purpose, accuracy and known risks.
Human Oversight	Designing the system so that human intervention and supervision are always possible, preventing blind automation.
Registration	High-risk systems must be registered in a public European database.
Conformity Assessment	The system must undergo a conformity assessment procedure, often involving a notified body (third-party certification).

The Future of Manufacturing and Compliance

For manufacturing companies, the challenge is twofold: to innovate with AI while maintaining a high level of safety and, at the same time, to ensure full compliance with the AI Act. Not all AI solutions in the factory (such as simple predictive stock analysis) will be high risk, but a careful mapping and risk assessment (AI Risk Assessment) of every tool used is essential in order to avoid heavy penalties.

This new regulatory framework pushes companies to build in ethics and transparency from the design stage (security by design and privacy by design), ensuring that AI is an engine of efficiency that fully respects people's safety and rights.