It’s a completed transaction. The EU’s Artificial Intelligence Law is enacted. The European Parliament adopted the latest legislation by an overwhelming majority on 14 June 2023.
The AI Law, introduced in April 2021, aims to tightly regulate AI services and mitigate the risks posed by AI. The first draft includes measures such as exploiting biometric data, mass surveillance systems, and adding safeguards to enforcement algorithms, anticipating the surge in adoption of generative AI tools beginning in late 2022. bottom.
The latest draft, introduced in May 2023, introduced new measures to manage the “basic model”.
These include a stepwise approach to AI models ranging from ‘low and minimal risk’ to ‘limited risk’, ‘high risk’ and ‘unacceptable risk’ AI practices.
“Low and minimal risk” AI tools will not be regulated, but “limited risk” AI tools will require transparency. However, “risky” AI practices will be heavily regulated. The EU will require a database describing where, when and how general-purpose and high-risk AI systems are deployed within the EU.
“This database should be freely and publicly accessible, easily understood, and machine-readable. It should also be user-friendly and easily navigated, allowing the general public to There should be minimal search capabilities to search the database for , location and risk categories. [and] Keywords,” the law says.
AI models with “unacceptable risks” are completely banned.
Similar to the General Data Protection Regulation (GDPR) for personal data protection, the AI law will also be the first AI law in the world to impose hefty fines of up to €30 million ($32 million) for violations. 6% of world profits.
Edward Machin, senior attorney for the data, privacy and cybersecurity team at law firm Ropes & Gray, welcomed the bill, saying: “We can think of risk uses beyond chatbots, such as facial recognition technology and profiling systems. prize.”
UK: innovation over regulation
With this pioneering regulation, EU lawmakers hope other countries will follow suit. In April, 12 EU lawmakers working on the AI bill called for a world summit to find ways to manage the development of advanced AI systems.
Several other countries, such as Canada and its AI & Data Act, have begun to work on similar regulations, but the US and UK appear to be taking a more cautious approach to regulating AI practices.
The UK government said in March that it was taking an “innovation-promoting approach to AI regulation”. The company released a white paper outlining its plans, in which it said there would be no new laws or regulatory bodies for AI. Instead, responsibility will shift to existing regulators in areas where AI is applied.
In April, the UK announced a £100 million ($125 million) investment to launch a Foundation Model Task Force. It is hoped that this will help promote the development of AI systems that will boost the country’s GDP.
On June 7, UK Prime Minister Rishi Sunak announced that the UK would host the first global AI summit in the fall of 2023.
Then, on June 12, Sunak announced at London Tech Week that Google DeepMind, OpenAI and Anthropic had agreed to release their AI models to the UK government for research and safety purposes.
Machin commented: “It remains to be seen whether the UK will reconsider its light-weight approach to regulation as public concern about AI grows, but in any case the AI law will continue to influence parliamentarians in Europe and beyond.” Foreseeable For the future. ”
UK National Cyber Security Center (NCSC) CEO Lindy Cameron addresses the UK’s leading role in AI development during a keynote address at Chatham House’s Cyber 2023 conference on June 14. Mentioned.
“As the world leader in AI, third only to the United States and China,” she said. […] The UK is well positioned to take advantage of artificial intelligence developments safely. So the Prime Minister’s AI Summit comes at a perfect time to bring together the world’s experts to share their ideas. “
She outlined three goals for NCSC to address the cyber threats posed by generative AI. It helps organizations understand the risks, maximize the benefits of AI to the cyber defense community, and understand how adversaries attack. […] Using AI and how it can be disrupted – She didn’t mention AI regulation.
