The EU AI Act sorts AI systems into risk categories. This article explains how the Eye-Able software solutions fall within this classification.
No prohibited AI practices
The Eye-Able software, including the Ally assistant, uses no AI practices prohibited under Art. 5 of the AI Act. In particular, it contains no systems for the subliminal influencing of people, for exploiting weaknesses or vulnerability, for biometric categorization or remote identification, for social scoring, or for manipulating human behavior.
Not a high-risk AI system
The Eye-Able software is not a high-risk AI system within the meaning of Art. 6 of the AI Act. The areas exhaustively listed in Annex III — including critical infrastructure, education, employment, essential private and public services, law enforcement, border control, the administration of justice, and democratic processes — are not applicable.
The software serves solely as technical support for creating and improving digital accessibility. It makes no automated decisions with legal or similarly significant effects on natural persons.
General-purpose AI model (GPAI)
Insofar as the Ally function is based on large language models, it incorporates a general-purpose AI model (GPAI) under Art. 3 no. 63 of the AI Act. This underlying model is provided by the model operator (e.g. the LLM vendor), not by Eye-Able. The obligations under Chapter V of the AI Act apply solely to the provider of that underlying GPAI model — here the model operator. Eye-Able, as provider of the AI system that integrates this model, has separate obligations under Art. 50 and the Art. 6/Annex III risk assessment, not under Chapter V.
Further articles
In the following articles, you’ll find further information on similar topics:
- The EU AI Act and the Eye-Able software solutions
- Labeling obligations under Art. 50 of the AI Act for deployers
If you have any further questions, our team is happy to help! In the following article, you’ll find ways to get in touch with us: