By Porter Anderson, Editor-in-Chief | @Porter_Anderson
Trialogue Begins in Negotiating the BillIn Brussels, the Federation of European Publishers today (June 14) has issued a statement, welcoming the European Parliament’s approval of rules for “artificial intelligence,” legislation known as Europe’s AI Act.
As reported by Ryan Browne for CNBC, “During a critical Wednesday vote, the parliament adopted the AI Act with 499 votes in favor, 28 against, and 93 abstentions. The regulation is far from becoming law, but it is likely to be one of the first formal rules for the technology globally.”
At The New York Times, Adam Satariano calls the AI Act “a potential model for policymakers around the world as the grapple with how to put guardrails on the rapidly developing technology.”
Cat Zakrzewski and Cristiano Lima at the Washington Post write, “EU officials are moving much faster than their US counterparts, where discussions about AI have dragged on in Congress despite apocalyptic warnings from even some industry officials.”
The publishers’ federation writes in its statement, “While AI offers great opportunities, including for the publishing industry, it also raises serious concerns, particularly in the field of copyright and transparency. The AI Act marks the first attempt in the world to define dedicated rules for AI, to ensure that this technology can develop without harming society.
The federation “particularly welcomes the decision of the European Parliament to tackle the rising issue of generative AI, such as ChatGPT, which has gained significant attention [in] the past few months. The developers of this type of AI use vast quantities of content, including copyright-protected books and articles, to train their applications and allow them to generate ‘new’ content. The copyright-protected content used for the training is often accessed and used without the knowledge or consent of their rights holders, and even through illegal websites.
The federation goes on to write:
“Generative AI is a black-box that does not allow rights holders to know whether their works have been used without authorization, even if they exercised their right to opt-out. The rights and future works of authors are also put at risk by AI-generated content mimicking their style. Finally, consumers could be misled into thinking that an AI-generated content is a genuine creative work from a human author.
“The Parliament recognized the seriousness of the issue by creating specific obligations for generative AI and underlining the requirement to respect copyright, which includes the possibility for rights holders to opt-out from text and data mining under the 2019 DSM Directive. While this is a positive first step, further improvements are necessary to ensure training dataset transparency, as a mere ‘detailed summary’ would not allow rights holders to protect their rights.”
Federation president Ricardo Franco Levi is quoted, saying, “Today, the European Parliament showed the way by deciding to tackle the challenge of generative AI.
“Now we call on the co-legislators to show ambition in the trialogue negotiations to prevent any risk of ‘data laundering.’
“Providers of generative AI must grant access to the detailed list of books and articles they used for training and information about where they were collected. Without meaningful transparency, our rights will become purely virtual, to the sole benefit of big technological players and to the detriment cultural diversity and society.”
As Lisa O’Carroll in Strasbourg describes it for today’s piece in the Guardian, “To combat the high risk of copyright infringement, the legislation will oblige developers of AI chatbots to publish all the works of scientists, musicians, illustrators, photographers, and journalists used to train them.
“They will also have to prove that everything they did to train the machine complied with the law.”
Prohibiting ‘Risks to People’s Safety’
The European Parliament’s news service today informs journalists that members of parliament have expanded a list of prohibited AI systems that would carry what the legislation deems to be an “unacceptable level of risk to people’s safety, such as those used for social scoring (classifying people based on their social behavior or personal characteristics.)”
More bans on intrusive and discriminatory uses called for in the European Parliament’s information about the act are:
- Real-time remote biometric identification systems in publicly accessible spaces
- Post remote biometric identification systems, with the only exception of law enforcement for the prosecution of serious crimes and only after judicial authorization
- Biometric categorization systems using sensitive characteristics (e.g. gender, race, ethnicity, citizenship status, religion, political orientation)
- Predictive policing systems (based on profiling, location or past criminal behavior)
- Emotion recognition systems in law enforcement, border management, the workplace, and educational institutions
- Untargeted scraping of facial images from the Internet or CCTV footage to create facial recognition databases (violating human rights and right to privacy)
Among obligations specified in the act, “Generative AI systems based on … models like ChatGPT would have to comply with transparency requirements (disclosing that the content was AI-generated, also helping distinguish so-called deep-fake images from real ones) and ensure safeguards against generating illegal content. Detailed summaries of the copyrighted data used for their training would also have to be made publicly available.”
Significantly, AI systems used “to influence voters in elections” are considered by the legislation “to be high-risk,” and thus not acceptable.
Following today’s vote, the European Union’s three branches (the parliament, the European Commission, and the Council of the European Union) are to go into “trialogue,” in which negotiations are used to construct a final version of the AI Act, with many hoping this can be accomplished by the end of the year.
More from Publishing Perspectives on the European Parliament is here, more from us on AI, ‘artificial intelligence,’ is here, more on the Federation of European Publishers is here, and more on Europe and its publishing interests is here.