EHIMA calls for clear and proportionate guidance under the EU AI Act
Artificial intelligence is helping modern hearing aids and hearing implants deliver clearer speech, reduce background noise and adapt more effectively to individual users and changing listening environments.
These functions can significantly improve hearing performance and user experience. However, their use in a medical device should not automatically result in the AI system being classified as high-risk under the EU AI Act.
In its response to the European Commission’s consultation on the draft guidelines for the classification of high-risk AI systems, EHIMA calls for a clear, functional and technology-neutral interpretation of the rules.
Not every AI function in a medical device is a safety component
Under Article 6(1) of the AI Act, an AI system used in a product covered by EU product-safety legislation may be classified as high-risk if it is itself a regulated product or performs the function of a safety component.
The decisive question should therefore be whether the AI system performs a safety function or whether its failure could endanger the health and safety of people or property.
The mere fact that an AI system is incorporated into a medical device regulated under the Medical Devices Regulation should not be sufficient to classify it as high-risk.
Modern hearing aids may use machine learning or deep neural networks for functions such as:
- speech enhancement and background-noise reduction;
- classification of listening environments;
- detection of the user’s listening intentions; and
- personalised amplification and fitting.
These functions are primarily intended to improve performance, comfort and the user experience. If they fail or malfunction, the usual consequence would be reduced hearing benefit, poorer sound quality or inconvenience—not a threat to health or safety.
Such AI functions should therefore generally not be considered safety components within the meaning of the AI Act.
The same function should lead to the same regulatory outcome
EHIMA also calls for the guidelines to apply the concept of a safety component consistently to both hardware and software.
Standalone medical software can contain several modules, algorithms and functional components, just as a physical product can contain different hardware components. Some may perform safety-related functions, while others may simply optimise performance or improve usability.
The regulatory classification should therefore depend on the function performed and the risks associated with that function—not on whether it is implemented in hardware, embedded software or a standalone software product.
Equivalent functions should lead to equivalent regulatory outcomes. Otherwise, classification could depend on technical design choices rather than on the actual level of risk.
Clear examples are needed
EHIMA recommends that the final guidelines include a specific example of an AI-enabled hearing aid or hearing system whose AI functionality is limited to speech enhancement, noise reduction, environmental classification or personalised sound processing.
The example should clarify that such an AI system is not high-risk merely because the hearing device itself is subject to the Medical Devices Regulation or requires third-party conformity assessment.
EHIMA calls on the European Commission to ensure that the final guidelines:
- distinguish clearly between safety-related and performance-enhancing AI functions;
- confirm that medical-device status does not automatically make incorporated AI high-risk; and
- apply a consistent, technology-neutral approach to hardware and software.
A proportionate, risk-based interpretation will provide legal certainty, avoid unnecessary regulatory burdens and support the continued development of innovative hearing technologies—without compromising the high level of safety expected from medical devices in Europe.
