The executive branch of the EU is expected to lay out the proposal in draft AI rules to be announced next week. In order for the rules to become EU law, they must first be scrutinised and passed by the European Parliament.

Under the proposals, companies which use automated tools to surveil people and their behaviour could be fined up to four per cent of their annual turnover; this maximum fine is the same as that which may be applied for data privacy violations under GDPR.

The rules would ban AI applications used for manipulating human behaviour and social scoring. These proposals come amid criticism of AI software used for recruitment, approving financial loans and other important decisions, which can reinforce existing biases.

The EU fines could be applied for developing and selling banned AI apps; providing false information to authorities, or failing to comply with investigations.

According to documents seen by Reuters, the rules would also include safeguards and prior authorisation for the use of AI in applications considered “high risk”, such as those with implications for health, safety and fundamental rights and freedoms.

AI applications used for remote biometric ID (e.g. facial recognition used in public spaces); job recruitment; access to educational institutions; assessing creditworthiness and asylum, and visa applications all fall under this category.

These systems would require human oversight. The draft proposes the establishment of national bodies to issue five-year certificates for use of the most high-risk systems.

The document warned of the implications of mass automated surveillance for human rights and dignity, democracy and freedom. Some governments and companies have attracted criticism for their intrusive and often opaque use of mass AI surveillance to track the behaviour of individuals.

“The use of [AI] for the purposes of indiscriminate surveillance of natural persons should be prohibited when applied in a generalised manner to all persons without differentiation,” the document said. These methods could include “monitoring and tracking of natural persons in digital or physical environments” and “automated aggregation and analysis of personal data from various sources”. Exceptions would be permitted for the purposes of public security and defence.

The proposals will be presented next week by Margrethe Vestager, who has been appointed the European Commission’s digital lead following a groundbreaking term as Commissioner for Competition during which she investigated and held to account Big Tech companies including Google, Apple, Facebook and Amazon.

The rules would establish a pioneering regulatory framework for a broad, ubiquitous and often-controversial family of technologies often considered to be developing too rapidly for policymakers and regulators to keep up. Regarding this problem, the Commission has also proposed the establishment of a ‘European Artificial Intelligence Board’, composed of a representative from each of the EU member states, a representative of the Commission, and the European Data Protection Supervisor. This panel would be tasked with issuing recommendations to the Commission regarding prohibited AI practices and the list of high-risk AI systems.

The executive branch of the EU is expected to lay out the proposal in draft AI rules to be announced next week. In order for the rules to become EU law, they must first be scrutinised and passed by the European Parliament.

Under the proposals, companies which use automated tools to surveil people and their behaviour could be fined up to four per cent of their annual turnover; this maximum fine is the same as that which may be applied for data privacy violations under GDPR.

The rules would ban AI applications used for manipulating human behaviour and social scoring. These proposals come amid criticism of AI software used for recruitment, approving financial loans and other important decisions, which can reinforce existing biases.

The EU fines could be applied for developing and selling banned AI apps; providing false information to authorities, or failing to comply with investigations.

According to documents seen by Reuters, the rules would also include safeguards and prior authorisation for the use of AI in applications considered “high risk”, such as those with implications for health, safety and fundamental rights and freedoms.

AI applications used for remote biometric ID (e.g. facial recognition used in public spaces); job recruitment; access to educational institutions; assessing creditworthiness and asylum, and visa applications all fall under this category.

These systems would require human oversight. The draft proposes the establishment of national bodies to issue five-year certificates for use of the most high-risk systems.

The document warned of the implications of mass automated surveillance for human rights and dignity, democracy and freedom. Some governments and companies have attracted criticism for their intrusive and often opaque use of mass AI surveillance to track the behaviour of individuals.

“The use of [AI] for the purposes of indiscriminate surveillance of natural persons should be prohibited when applied in a generalised manner to all persons without differentiation,” the document said. These methods could include “monitoring and tracking of natural persons in digital or physical environments” and “automated aggregation and analysis of personal data from various sources”. Exceptions would be permitted for the purposes of public security and defence.

The proposals will be presented next week by Margrethe Vestager, who has been appointed the European Commission’s digital lead following a groundbreaking term as Commissioner for Competition during which she investigated and held to account Big Tech companies including Google, Apple, Facebook and Amazon.

The rules would establish a pioneering regulatory framework for a broad, ubiquitous and often-controversial family of technologies often considered to be developing too rapidly for policymakers and regulators to keep up. Regarding this problem, the Commission has also proposed the establishment of a ‘European Artificial Intelligence Board’, composed of a representative from each of the EU member states, a representative of the Commission, and the European Data Protection Supervisor. This panel would be tasked with issuing recommendations to the Commission regarding prohibited AI practices and the list of high-risk AI systems.