General-Purpose AI Model
AI model trained on broad data at scale that displays significant generality and can be integrated into many downstream systems.
Definition
Article 3(63) of the AI Act defines a GPAI model as an AI model that displays significant generality, is capable of competently performing a wide range of distinct tasks, and can be integrated into a variety of downstream systems or applications. Chapter V of the Act imposes obligations on GPAI providers, including technical documentation, downstream information, copyright policy, and a public training-data summary.What this means in practice
Foundation models used as components in medical AI (for example large language models supporting clinical documentation or vision-language models used in radiology copilots) sit upstream of the medical device. The device manufacturer remains responsible for the finished device, but the GPAI provider carries its own Chapter V duties that MedTech integrators should verify contractually.Examples
- A large language model, trained on diverse text datasets, can be integrated into a clinical decision support system for suggesting diagnoses and into a separate application for summarizing patient notes.
- A foundational vision model, capable of identifying various objects and patterns in images, is used by one MedTech company for preliminary anomaly detection in X-rays and by another for surgical tool recognition during procedures.
- A general-purpose AI system designed for data analysis and prediction, after fine-tuning, assists in optimizing hospital logistics and is also adapted to forecast disease outbreaks from public health data.
- •Assuming that using a third-party GPAI shifts device-manufacturer liability. It does not; the MedTech firm remains responsible for the intended use and clinical claim.
- •Treating GPAI documentation from an upstream provider as sufficient for MDR/IVDR technical documentation. It is a component input, not a substitute.
Frequently asked questions
Related terms
Grouped by themeEditor's picks
· Hand-selected related conceptsAI systems captured by Article 6 of the EU AI Act because they are safety components of regulated products or listed in Annex III.
EU regulation establishing risk-based requirements for AI systems, including most medical AI.
Large pretrained model adaptable to many downstream clinical tasks via fine-tuning or prompting.
More in Regulatory
· Same categoryFDA submission demonstrating a device is substantially equivalent to a legally marketed predicate.
Two alternative disclosure mechanisms in a 510(k) submission, a Summary is public on FDA's website; a Statement promises to share Safety & Effectiveness info on request.
Formal mechanism to ask FDA whether a product is a device and, if so, its likely classification.
510(k) variant that relies on FDA guidance, special controls, or recognized consensus standards.
Where this term appears across MedTech Terms.
Latest in MedTech
Sources
3 sourcesEvery citation below opens the original document. Each is graded against our source-tier hierarchy so you can see what rests on binding law versus commentary.
- 1Regulation (EU) 2024/1689, Article 3 and Chapter VTier 1 VerifiedEUR-Lexeur-lex.europa.eu
- 2AI OfficeTier 2 UncheckedEuropean Commissiondigital-strategy.ec.europa.eu
- 3European Commission - Medical DevicesTier 2 VerifiedEuropean Commissionhealth.ec.europa.eu
Inline markers like [1] jump to the matching reference above.