The EU AI Act is a cornerstone in regulating artificial intelligence (AI). The legislation shapes the ‘do’s and ‘don’ts’ for AI in the European Union. A fascinating aspect of this Act is Article 5, which outlines prohibited AI practices. Let’s delve into these prohibitions to understand the AI puzzle in the EU better.

Banned AI practices (Art. 5.1)

Article 5 sets the ground rules for AI usage in the EU. It casts a wide net over various prohibited practices, guiding us on what is unacceptable.

The no-go zone of subliminal techniques (Art. 5.1(a))

The Act makes it clear – AI systems that use sneaky, subconscious methods to change a person’s behaviour drastically are off-limits. These harmful methods can affect individuals physically or psychologically.

Safeguarding vulnerable groups (Art. 5.1(b))

The protection of vulnerable individuals is paramount. The Act prohibits AI systems designed to manipulate vulnerable individuals, such as children or people with disabilities, by significantly altering their behaviour in harmful ways.

Restrictions on evaluating trustworthiness (Art. 5.1(c))

The Act curbs the overreach of AI in evaluating personal trustworthiness. Public authorities or entities acting on their behalf cannot use AI systems to assign ‘social scores’ based on an individual’s behaviour or characteristics. These scores should not lead to unfavourable or unjust treatment in unrelated social contexts.

The rules of engagement for biometric identification (Art. 5.1(d))

‘Real-time’ remote biometric identification systems have their specific boundaries. They can only be deployed in public spaces for law enforcement when it’s absolutely necessary to prevent serious harm, search for potential victims or track down significant criminal offenders.

Considerations for biometric identification use (Art. 5.2)

Using real-time biometric identification systems must be a balanced act. The Act urges law enforcement to consider the seriousness and likelihood of harm that might occur from not using the system against the potential impact on people’s rights and freedoms.

Getting the green light for biometric system use (Art. 5.3)

Each use of these powerful biometric systems requires prior permission. The systems can be used without initial authorisation in exceptional circumstances requiring urgent action. The granting authority must be convinced of the necessity and proportionality of using the system for a specific purpose.

Member state’s say in biometric system use (Art. 5.4)

Member States can decide to permit the use of real-time biometric identification systems within their borders. Their national laws must outline the rules and conditions under which these systems can be used, supervised, and authorised.

Framework of EU AI Act prohibited practices

Prohibited AI practices under the EU AI Act create a clear framework for what AI can and cannot do within the EU. It safeguards individual rights and freedoms while recognising the value of AI in specific law enforcement situations.

Actions you can take next

  • Stay informed about AI regulations in your country and the EU context by regularly reviewing updates to the EU AI Act. You can read the full act or join our mailing list.
  • Consider how these regulations impact your personal or professional use of AI technologies. With knowledge comes the power to navigate the complex AI regulatory terrain effectively.