EU Passes Law to Restrict High-Risk AI Uses

European Union passes law to regulate AI applications

The European Union passed the first comprehensive law in the world regulating artificial intelligence. The historical AI Act is expected to officially become law by May or June 2024 pending formal approval from EU member countries, with provisions taking effect six months after the inclusion of the rule in the lawbooks.

Rules for general-purpose AI systems such as ChatGPT will be implemented a year after the law goes into force. The entire set of regulations is expected to be in effect by mid-2026.

It is an attempt to protect fundamental rights, democracy, the rule of law, and environmental sustainability from high-risk AI, while also fostering innovation and establishing Europe as a leader in the field. The rule assigns obligations to AI based on its potential risks and level of impact.

The new rule bans certain AI applications that have unacceptably high levels of risk and threaten the rights of the citizen including biometric categorization systems that are used to detect identification of a person or sexual orientation. Emotion recognition in the workplace and schools, predictive policing, and artificial intelligence that manipulates human behavior or exploits people's vulnerabilities would also be banned. AI systems that involve high-risk applications used for the hiring process or law enforcement will be more strictly regulated with developers expected to demonstrate that their models are safe, transparent, and comply with privacy standards.

There will be little to no regulation on lower-risk AI tools but still, developers will be required to label AI-generated deepfake images, video, or audio of existing persons, places, or events as artificially modified. The law applies to models operating in the EU, and any organization that violates the rules will need to give a fine of up to 7% of its annual global earnings.

Enforcing the law, each country of the EU will set up its own AI watchdog and the citizens will have a right to submit complaints about AI tools in case their rights get violated. Brussels will set up a stand-alone AI office that will enforce and supervise the law for the general-purpose AI systems. There will

The biggest and most powerful AI models that the EU considers to pose "systemic risks," such as OpenAI's GPT4 and Google's Gemini, will be scrutinized. Companies that provide these systems must assess and mitigate risks, implement cybersecurity measures, report any severe incidents resulting from their systems, and disclose how much energy their models use. All general-purpose AI systems must develop a policy that ensures the content used to train their models complies with European copyright law.

Regulatory sandboxes and real-world testing will have to be established at the national level and made accessible to small and medium-sized enterprises and start-ups to develop and train innovative AI before it is launched in the market.

Previously, the Digital Services Act, which targeted social media abuse, and the Digital Markets Act, which went into effect on March 7. The Digital Market Act aims to combat market dominance by so-called digital "gatekeepers," and the EU's AI act goal is to be the global default legislation.