Which artificial intelligence systems are banned by the AI Act

Introduction

Within the regulatory model introduced by the European regulation on artificial intelligence, there is a category of completely prohibited systems.

These are applications classified as unacceptable risk.

According to the Artificial Intelligence Act, these systems present such a high level of risk to people's fundamental rights that it cannot be mitigated through simple technical or procedural requirements.

For this reason, the European legislator has chosen to directly prohibit their use.

What is meant by unacceptable risk

An AI system is considered to be at unacceptable risk when its use can compromise fundamental rights, individual freedoms or democratic principles.

In these cases, the regulation does not introduce compliance obligations or additional controls.

The use of such systems is simply prohibited within the European Union, except for some very limited exceptions provided for specific contexts.

Some examples of prohibited practices

Among the prohibited applications we find various practices considered particularly invasive or manipulative.

For example:

systems designed to manipulate people's behavior by exploiting psychological vulnerabilities

social scoring systems that classify citizens based on their behavior or personal characteristics

some forms of real-time biometric identification in public spaces.

These technologies are considered incompatible with the protection of fundamental rights and with the principles of a democratic society.

Conclusion

The prohibitions provided by the AI Act represent the highest level of protection introduced by the regulation.

By clearly defining which applications are not acceptable, the European Union establishes the limits within which the development and use of artificial intelligence can take place responsibly.