Sun Apr 18 2021
New proposals would also ban any algorithm designed to judge a person’s ‘trustworthiness’ based on their social behaviour
The European Union is all set to outlaw almost all uses of AI systems that have been programmed for ‘indiscriminate surveillance’ or for the ranking of a persons ‘social behaviour’.
The new rules are currently being drafted for legislation that will be forward by the European Commission in the next few weeks.
Early copies of the draft state that organisations operating within the EU won’t be allowed to use AI systems that are designed to track a person in physical environments or also aggregate data from other sources. Algorithms that make it possible to measure a person’s ‘trustworthiness’ based on social pattens or personality traits will also be banned.
Finally, the new legislation will ban any system that is designed to exploit data about groups of people as a whole.
Special dispensation will also be required for the deployment of biometric identification systems in all public spaces, with all ‘high-risk’ AI required to submit to a through review before any organisation is allowed to use it.
Examples of ‘high-risk’ AI include, but won’t be limited to, recruitment algorithms, crime-predicting algorithms, any system used to determine an individual’s credit worthiness and any system used that might establish priority for the dispatch of emergency services.
For all of the above a member state would be required to appoint an assessment body to ensure such systems are trained on unbiased data, with adequate human oversight.
Very similar to GDPR, any organisation violating these rules, whether intentionally or not, runs the risk of being fined 4% of their annual global revenue.
It’s thought however that the legislation will include some exemptions, notably AI used for military purposes or any designed to safeguard public security.
Some of the uses and applications of artificial intelligence may generate risks and cause harm to interests and rights that are protected by Union law. Such harm might be material or immaterial, insofar as it relates to the safety and health of persons, their property or other individual fundamental rights and interests protected by Union law.
As well as limiting the use of AI, the draft legislation also calls for the formation of an overarching European AI Board that can help the European Commission decide which AI systems are in fact, ‘high-risk’.
The board would have a representative from every EU member state, one from the European Commission as well as a European Data Protection Supervisor.
Sun Apr 18 2021