It is an understatement to say that their proposals will be scrutinized closely by the Tech giants. MEPs want to adopt fairly strong measures to try to regulate artificial intelligence technologies.
This project has been in discussion for quite a while now, but it now incorporates measures relating directly to ChatGPT, and its rivals such as Bard and Bing Chat. Here are 4 measures that could change the future of AI in Europe.
The major issue of copyright
This is one of the most glaring issues with generative language models. The AI relies on writing or images to produce its content, and these may be subject to copyright. Lawsuits have already been launched.
MEPs are aware of this issue, and they foresee that the developers of these applications will have to declare whether copyrighted material was used to train their AIs. Enough to allow creators to demand payment if their material has been used.
What responsibility for companies?
ChatGPT and its rivals can enable extremely beneficial applications and bring progress. Unfortunately, we also know that they can also fall into the wrong hands and be the subject of malicious uses, in particular to carry out scams or cyberattacks.
Parliamentarians therefore want the responsibility for the misuse of these AIs to lie with the developers, and not with the smaller companies that will use them.
Greater transparency
According to the proposals that our colleagues from the Financial Times, MEPs want chatbots to make it clear to their users that they are not human. Transparency must be in order, and we imagine that here again, abusive uses of ChatGPT are in the sights.
Predictive policing banned
This measure does not directly concern language models, but it is the subject of heated discussions. Elected officials insisted that the text include a ban on predictive policing. This applies to the use of algorithms to identify a suspect or prevent a crime.
It will also be necessary to see whether or not this decision will come into contradiction with the recent French law which validated the use of algorithmic video surveillance (VSA) in order to secure the Paris 2024 Olympic Games.
Nothing says that all these ideas will be retained in the final text which remains under discussion. It should be finalized before the end of 2023 or the beginning of 2024 and will then be translated into national laws a little later.
