EU AI Act Moves Forward

The EU Parliament Wednesday  passed the AI Act, moving it one step closer to enactment.

The AI Act seeks to provide clear guidelines and restrictions on AI usage, with a focus on mitigating risks posed by certain AI applications.

Key points in the regulation include:

1. Restrictions on Surveillance: The proposed legislation prohibits the use of real-time remote biometric surveillance in public spaces, and targets the creation of facial recognition databases through harvesting surveillance footage or scraping the internet.

2. Ban on Predictive Policing Systems: The draft also recommends a ban on predictive policing systems that use data analysis of past criminal behavior to predict future illicit activities.

3. Disclosure of AI-generated Content: In an important development, the EU lawmakers have proposed requiring AI systems like ChatGPT to disclose when content has been generated by AI.

4. Requirement for Generative AI Systems: Companies using generative tools will need to disclose copyrighted material used for training the systems. Additionally, these companies will need to design their AI models to prevent illegal content creation.

5. Risk Assessment: Companies involved in the development of "high-risk applications" will be required to carry out a fundamental rights impact assessment and environmental impact evaluation.

6. Penalties: Noncompliance with the regulations could result in substantial fines, with proposed penalties up to 6% or 7% of a company's global revenue.

While there is broad agreement on the need for some level of regulation, the AI industry response has been mixed. Some stakeholders fear these regulations could stifle innovation, while others have voiced their support.

 The EU Parliament, European Commission, and EU member states will now hold further discussions to finalize the Act, with the aim of reaching an agreement by the end of the year.