Progress is being made in the European Union Parliament regarding the proposed regulation for Artificial Intelligence (AI). Negotiators from the various factions met on Thursday and reportedly agreed upon a package including tighter rules for AI systems that are used for multiple purposes, such as speech or image recognition (“General Purpose AI”). Developers of products such as OpenAI’s ChatGPT will be required to indicate if they have used copyrighted material to train their models, according to the Financial Times. This will enable authors and publishers to demand for licensing fees or royalties.
Creators and organizations which earn money through intellectual creations are increasingly affected by the competition posed by AI. The operators of the generators train their models with millions of pictures and texts found on the internet, without asking the authors and publishers for permission to use them. Legal proceedings have already been launched against British company Stability AI and other manufacturers in the US and UK. Now, EU legislators should also make sure that the developers are responsible for the misuse of AI programs, not businesses and consumers.
Liberals, Greens, the Left and parts of the Social Democrats basically agree that predictive policing, social scoring and mass biometric surveillance in public spaces such as automated facial recognition should be prohibited by the EU. Nevertheless, a post-hoc biometric assessment for purposes like criminal investigations will remain permissible, as advocated for by the German government, despite claims that it violates the coalition agreement. One of the most contested cases in recent years was the use of facial recognition by Hamburg Police in search for rioters during the G20 summit.
The agreement however is shaky, as reported by Euractiv. Members of the European People’s Party (EPP) – which includes CDU & CSU – bear within it a ban on real-time biometric identification. It is uncertain however, whether the EPP leadership will challenge the proposal with an alternative one as the final parliamentary vote approaches. If not approved, the Conservative faction would likely vote against all progress made. The EU states advised in December that the exception catalogue proposed by the EU Commission on biometric surveiliance should be expanded.
Euractiv reports that the Free Democratic Party’s (FDP) proposal of a prohibition of AI systems “for the general surveillance, recognition, and evaluation of private material in interpersonal communication services” is also cause for debates. The Liberals want to prevent any actions which undermine the end-to-end encryption, including chat control driven by the Commission. EPP is opposed to this as well. In case their motion does not pass, Greens and the Left petition for a ban on emotion and behaviour recognition for areas like law enforcement, border control, workplace, and education. The Committee on Internal Market and Internal Security will probably define the Parliament’s proposed action come 26 April.
ChatGPT is OpenAI’s cutting-edge natural language processing model that enables machines to generate long-form, conversation-based text responses. It is currently being used by some of the major companies in the world for storytelling, customer support and more. OpenAI was founded in late 2015 by technology entrepreneurs Elon Musk and Sam Altman, with the mission to make sure AI technology is beneficial to humanity.
Sam Altman is an American computer scientist and entrepreneur, most noted for his roles as the president of Y Combinator; a director of companies such as OpenAI, Loopt, Mixer Labs, and Ifunded; and a columnist for techcrunch. Altman holds degrees from Stanford University in symbolic systems and computer science and is the author of several books, including The Animation of Data Abstraction. He is also involved in various philanthropies, such as the Future of Humanity Institute, which researches artificial intelligence and ethical implications of biotechnologies.
