The \"Artificial Intelligence ACT\" or \"Artificial Intelligence Regulation\" will soon be in force, protecting EU citizens from the potential excesses of AI while giving the EU a major role in the global AI industry.<\/p>\n
After several months of negotiations a political agreement<\/u>he European co-legislators (Parliament and Council) reached an agreement on 9 December 2023, paving the way for a final vote.<\/p>\n
Here's what the President of the European Commission, Ursula von der Leyen, has to say about it:<\/p>\n
\u00abArtificial intelligence is already changing our daily lives. And this is only the beginning. AI, when used wisely and on a broad scale, promises considerable benefits for our economy and society. That is why I am delighted with the political agreement on the regulation on artificial intelligence reached today by the European Parliament and the Council. This European regulation is the world\u2019s first comprehensive legal framework for artificial intelligence. It is therefore a historic moment. The AI Regulation brings European values into a new era. By focusing regulation on identifiable risks, the agreement reached today will foster responsible innovation in Europe. By safeguarding the security and fundamental rights of citizens and businesses, the AI Regulation will contribute to the development, deployment and adoption of trustworthy AI in the EU. Our AI Regulation will make a substantial contribution to the development of global rules and principles for human-centred AI.\u00bb<\/p>\n
For the past four years, the European Forum of Freemasons has had a working group accredited to the Commission and Parliament to contribute to the development of the future AI-ACT.<\/p>\n
Here are the main stages in the legislative process:<\/p>\n
Following the publication of the European strategy on artificial intelligence in 2018 and after extensive consultation with stakeholders, the High-Level Expert Group on Artificial Intelligence drew up guidelines for trustworthy AI in 2019, as well as an assessment checklist for trustworthy AI in 2020. At the same time, the first coordinated plan on AI was published in December 2018 as part of a joint commitment with the Member States.<\/p>\n
The Commission\u2019s White Paper on AI, published in 2020, set out a clear vision for AI in Europe: an ecosystem of excellence and trust, which lays the foundations for the political agreement reached today. The public consultation on the White Paper on AI attracted numerous contributions from around the world. The White Paper was accompanied by a \u00abreport on the implications of artificial intelligence, the Internet of Things and robotics for safety and liability\u00bb, which concluded that current product safety legislation has a number of gaps that need to be addressed, particularly in the \u00abMachinery\u00bb Directive.<\/p>\n
The guiding principles of AI-ACT are as follows:<\/p>\n
The new rules, based on a future-proof definition of AI, will apply directly and uniformly across all Member States. They follow a risk-based approach:<\/p>\n
Minimal risk <\/u>: the vast majority of AI systems fall into this category. Low-risk applications, such as AI-based recommendation systems or spam filters, will be exempt from any obligations as they pose little or no risk to citizens\u2019 rights or safety. Companies may, however, voluntarily commit to adopting additional codes of conduct for these AI systems.<\/p>\n
High risk <\/u>: AI systems deemed to be high-risk will have to comply with strict requirements, particularly with regard to risk mitigation systems; the quality of the datasets used; the logging of activities; detailed documentation; the provision of clear information to users; human oversight; and a high standard of robustness, accuracy and cybersecurity. Regulatory sandboxes will facilitate responsible innovation and the development of AI systems that comply with these requirements.<\/p>\n
High-risk AI systems include, for example, certain critical infrastructure in the water, gas and electricity sectors; medical devices; systems that determine access to educational institutions or are used for recruitment purposes, as well as certain systems employed in the fields of law enforcement, border control, the administration of justice and democratic processes. Furthermore, biometric systems for identification, categorisation and emotion recognition are also considered to pose a high risk.<\/p>\n
Unacceptable risk <\/u>: AI systems deemed to pose a clear threat to people\u2019s fundamental rights will be banned. These include, in particular, AI systems or applications that manipulate human behaviour to deprive users of their free will, such as toys that use voice assistants to encourage minors to engage in dangerous behaviour, or systems that enable social scoring by states or companies, and certain predictive policing applications. Furthermore, certain uses of biometric systems will be prohibited, for example emotion recognition systems used in the workplace and certain systems for categorising individuals or for real-time remote biometric identification for law enforcement purposes in public spaces.<\/p>\n
Specific transparency obligations :<\/u> When using AI systems such as chatbots, users should be aware that they are interacting with a machine. Ultra-realistic video effects and other AI-generated content must be labelled as such, and users must be informed of the use of biometric systems for categorisation or emotion recognition. Furthermore, providers will be required to design systems in such a way that synthetic images, text, audio and video content are labelled in a machine-readable format and that their nature as artificially generated or manipulated content is detectable.<\/p>\n
The AI Regulation introduces specific rules applicable to general-purpose AI models<\/u>, which will ensure transparency throughout the value chain. For highly powerful models that may pose systemic risks, additional binding obligations will be imposed regarding risk management and the monitoring of serious incidents, the carrying out of model assessments and adversarial testing (adversarial testing). These new obligations will be implemented through codes of practice drawn up by the industry, the scientific community, civil society and other stakeholders, in collaboration with the Commission.<\/p>\n
With regard to governance<\/u>, the relevant national market surveillance authorities will oversee the implementation of the new rules at national level, whilst coordination at European level will be carried out by a newly established European AI Office within the European Commission. This new AI Office will also oversee the implementation of the new rules on general-purpose AI models and monitor compliance with them. Together with the national market surveillance authorities, the AI Office will be the first body in the world to enforce binding rules on AI; as such, it is expected to become an international benchmark. For general-purpose models, a scientific panel of independent experts will play a central role by issuing warnings on systemic risks and contributing to the classification and testing of models.<\/p>\n
Heavy fines are envisaged.<\/p>\n
Companies that fail to comply with the rules will be fined. The amount of the fines will vary depending on the offences committed: \u20ac35 million or 7 % of global annual turnover (whichever is higher) for breaches relating to prohibited AI applications; \u20ac15 million or 3 % of global annual turnover for breaches of other obligations set out in the regulation, and \u20ac7.5 million or 1.5 % of global annual turnover for providing inaccurate information. More proportionate upper limits are set for administrative fines to be imposed on SMEs and start-ups in the event of non-compliance with the AI Regulation.<\/p>\n
Next steps.<\/p>\n
The political agreement, which must now be formally approved by the European Parliament and the Council, will enter into force 20 days after its publication in the Official Journal. The AI Regulation will then come into application two years after it enters into force, with the exception of certain specific provisions: the bans will apply after six months, whilst the rules on general-purpose AI will apply after 12 months.<\/p>\n
To cover the transitional period leading up to the date on which the Regulation will apply generally, the Commission will launch an AI Pact. This will bring together AI developers from Europe and around the world who are committed to voluntarily implementing the key obligations set out in the AI Regulation ahead of the deadlines laid down therein.<\/p>\n
In order to promote rules on trustworthy AI at international level, the European Union will continue to work within forums such as the G7, the OECD, the Council of Europe, the G20 and the United Nations. Most recently, we endorsed the agreement reached by G7 leaders as part of the Hiroshima Process on Artificial Intelligence, which sets out international guiding principles and a voluntary code of conduct applicable to advanced AI systems.<\/p>\n
Jo\u00ebl De Decker<\/p>\n
References<\/u><\/p>\n
Various press releases published by the European Commission<\/p>\n
<\/p>","protected":false},"excerpt":{"rendered":"
The \"Artificial Intelligence ACT\" or \"Artificial Intelligence Regulation\" will soon be in force, protecting EU citizens from the potential excesses of AI while giving the EU a major role in the global AI industry. After several months of...<\/p>","protected":false},"author":2,"featured_media":873,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"footnotes":""},"categories":[9],"tags":[],"class_list":["post-872","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-info-lettres"],"acf":[],"_links":{"self":[{"href":"https:\/\/aepl.eu\/en\/wp-json\/wp\/v2\/posts\/872","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/aepl.eu\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/aepl.eu\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/aepl.eu\/en\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/aepl.eu\/en\/wp-json\/wp\/v2\/comments?post=872"}],"version-history":[{"count":0,"href":"https:\/\/aepl.eu\/en\/wp-json\/wp\/v2\/posts\/872\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/aepl.eu\/en\/wp-json\/wp\/v2\/media\/873"}],"wp:attachment":[{"href":"https:\/\/aepl.eu\/en\/wp-json\/wp\/v2\/media?parent=872"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/aepl.eu\/en\/wp-json\/wp\/v2\/categories?post=872"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/aepl.eu\/en\/wp-json\/wp\/v2\/tags?post=872"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}