The first prohibited practice concerns AI systems which use subliminal techniques beyond someones consciousness **or** purposefully manipulative or deceptive techniques. These systems have to have the goal or effect that they materially distort the behaviour of a person or group of persons by limiting their ability to make an informed decision. This limitation should cause a person to take a decision, which they would not have taken otherwise, in a manner that causes or is likely to case that person, another person or group of persons significant harm. [[Article 5]](1)(a) What does this prohibition mean? For more information we can look at [[Recital 29]]. In the Recital, subliminal components are described as audio, image, video stimuli which persons cant perceive as they are beyond human perception. Manipulative or deceptive techniques are only described by their effect together with subliminal components. These techniques subvert or impair a person's autonomy, decision-making or free choice in ways that people are not aware or which are able to deceive even when people are aware of them. An example of a context where this can occur is in machine-brain interfaces or virtual reality, since these contexts give more control over stimuli for an AI system. This prohibition is explicitly aimed at distortions of behaviour with significantly harmful effects (intentional or unintentional), not general behavioural changes. It extends the general prohibition of unfair competitive practices. [[Recital 29]] also notes that this prohibition is not aimed at lawful practices like medical treatments or common and legitimate commercial practices like advertising, as long as this is done in accordance with applicable law.