\n<\/p>\n
Following a round of intense negotiations this week, lawmakers in Brussels have now reached a \u201cprovisional agreement\u201d<\/a> on the European Union\u2019s proposed Artificial Intelligence Act (AI Act). The EU\u2019s AI Act<\/a> is anticipated to be the world\u2019s first comprehensive set of rules to govern AI and could serve as a benchmark for other regions looking to pass similar laws. <\/p>\n<\/div>\n According to the press release, negotiators established obligations for \u201chigh-impact\u201d general-purpose AI (GPAI) systems that meet certain benchmarks, like risk assessments, adversarial testing, incident reports, and more. It also mandates transparency by those systems that include creating technical documents and \u201cdetailed summaries about the content used for training\u201d \u2014 something companies like ChatGPT maker OpenAI<\/a> have refused to do so far.<\/p>\n<\/div>\n Another element is that citizens should have a right to launch complaints about AI systems and receive explanations about decisions on \u201chigh-risk\u201d systems that impact their rights. <\/p>\n<\/div>\n The press release didn\u2019t go into detail about how all that would work or what the benchmarks are, but it did note a framework for fines if companies break the rules. They vary based on the violation and size of the company and can range from 35 million euros or 7 percent of global revenue, to 7.5 million euros or 1.5 percent of global revenue of turnover.<\/p>\n<\/div>\n There are a number of applications where the use of AI is banned, like scraping facial images from CCTV footage, categorization based on \u201csensitive characteristics\u201d like race, sexual orientation, religion, or political beliefs, emotion recognition at work or school, or the creation of \u201csocial scoring\u201d systems. The last two banned bullet points are AI systems that \u201cmanipulate human behavior to circumvent their free will\u201d or \u201cexploit the vulnerabilities of people.\u201d The rules also include a list of safeguards and exemptions for law enforcement use of biometric systems, either in real-time or to search for evidence in recordings.<\/p>\n<\/div>\n It\u2019s expected that a final deal will be reached before the end of the year. Even then, the law likely won\u2019t come into force until 2025 at the earliest<\/a>.<\/p>\n<\/div>\n The first draft of the EU\u2019s AI Act was unveiled in 2021<\/a>, seeking to distinguish what actually counts as AI, and synchronize the rules for regulating AI technology across EU member states. That draft predated the introduction of fast-changing generative AI tools like ChatGPT<\/a> and Stable Diffusion, however, prompting numerous revisions<\/a> to the legislation.<\/p>\n<\/div>\n Further negotiations will still be required to finalize some details before the AI Act comes into force.<\/p>\n<\/div>\n<\/div>\n