Point 7(a) of Annex III classifies as high-risk AI systems intended to be used in the field of migration, asylum and border control managementas polygraphs or similar tools. Such AI systems are generally designed to analyse physiological, behavioural or biometric signals with the purpose of inferring deception by natural persons in migration, asylum or border-control settings. Where such systems are intended to be used by or on behalf of law enforcement authorities in those settings, they will be classified as high-risk pursuant to point 6(b) of Annex III.
The high-risk classification under point 7(a) of Annex III reflects the serious consequences that such systems may have for decisions on entry, stay or protection, combined with documented risks of inaccuracy and bias. Such systems are considered high-risk when intended to be used in the migration, asylum and border control contexts even if the competent authorities may not take the final decision on entry, stay or protection. Even where a human makes the final decision, the output of the system may significantly influence that decision due to overreliance and automation bias.
Polygraphs or similar tools cover a range of modalities, including analysis of voice patterns (voice-stress analysis), micro-expressions or facial muscle activity, gaze, posture, or body movements, thermal imaging of physiological responses, multimodal int erview platforms combining these signals (see chapter 3.6.3 Point 6(b): Polygraphs and or similar tools). If any of the modalities are used by the AI system for purposes other than to infer or detect deception (e.g. for language or dialect detection), the system is unlikely to be classified as high-risk under point 7(a) of Annex III.
a) Practical example of an AI system falling within the use case of point 7(a) of Annex
- Border guards deploy an AI system in secondary inspection at an external airport. During an officer-led interview, the system analyses the traveller's verbal and non-verbal responses and provides the officer with a supplementary credibility indicator.
- A consular authority uses an AI system that analyses an applicant's voice patterns and verbal responses during visa interviews conducted by consular officers, providing the officer with supplementary indicators relating to the veracity of the statements made.
Such systems fall within point 7(a) because they function as polygraph-type tools used in migration, asylum or border-control procedures to assess truthfulness. The filter mechanism of Article 6(3) AI Act does not apply to the systems described in the example above , since the system generates evaluative outputs that directly influence the substance of an individual examination and cannot be considered a narrow procedural, preparatory or post-decision task under the exceptions listed in that provision.
Additionally, such AI systems are not prohibited per se under Article 5(1)(b ) AI Act, unless designed or used to materially distort behaviour of vulnerable persons in a significantly harmful manner. Additionally, such AI systems are not prohibited under Article 5(1)(h) AI Act, nor are they classified as high-risk under point 1(a) of Annex IIIbecause they are not RBI systems: they operate with the person’s active involvement in a faceto-face interview (so not - ‘remote”), they do not perform one-to-many ide ntity matching against a reference database (they infer credibility, not identity), and their processing is not aimed at establishing identity (in real time).
b) Practical examples of AI systems falling outside of the use caseof point 7(a) of Annex
- AI system that processes interview audio (as lie-detection tools do) but does not infer deception. Instead, it automatically transcribes the interview or translates responses into another EU language for the case file.
- AI system intended only to create neutral, non-prioritised summaries of interviews for the file, without credibility scoring or recommendations. This system falls outside point 7(a) of Annex III because the system does not assess deception.
c) Practical examples of AI systems falling within the use case but exempted under the filter mechanism of Article 6(3)
- An AI system intended to normalise audio/video signals (e.g. denoising, adjusting lighting, extracting facial landmarks) so that caseworkers can review recordings more easily. If the system does not produce or imply deception inferences, it could benefit from the exception for AI systems intended to perform a preparatory task listed in Article6(3)(d) AI Act.