Yesterday, the European Parliament voted overwhelmingly to approve the EU AI Act. 523 MEPs voted in favour of the regulation, with 46 against and 49 abstentions. The landmark vote by the Parliament marks the end of a long process to introduce wide-ranging rules to govern AI.
Rules to govern AI on a EU-wide level were first proposed in 2021. Senior EU officials have argued that the new regulation's risk-based approach will protect citizens from technological risks, while also fostering innovation.
The EU's high-risk/tired approach to AI regulation:
- Unacceptable risk - existential threats that endanger the lives of citizens
- High risk - critical infrastructures that could put the life and health of citizens at risk, education or vocational training and safety components of products
- Limited risk - AI systems with risk of manipulation or deceit
- Minimal risk - AI systems that do not fall under the above-mentioned categories, but carry some level of risk
The new Act addresses the following areas:
- Manipulation
- Vulnerable groups
- Biometrics
- Social scoring
- Criminal profiling
- Facial recognition
The Act will enter into force on the 20th day following publication in the Official Journal of the European Union. It will be fully applicable after 2024 months.
Key dates:
- 6 months: prohibitions on unacceptable risk AI become applicable
- 12 months: obligations for general purpose AI governance become applicable
- 24 months: all AI Act rules become applicable, including obligations for high-risk systems
- 36 months: obligations for all other high-risk systems apply
How can UK businesses best prepare for the EU AI Act?
Join us on 21 March 3-4, for our free webinar: How UK businesses can best prepare for the EU AI Act
The webinar will help UK businesses to get a better understand the impact of the EU AI Act and will bring together legal and regulatory experts, along with senior industry players.