EU is Trying to Tame AI

In the 3 minutes you’re reading this, the human race will generate more data than in the entire history of the world from the dawn of time until the year 2000. A chat with friends, a search engine query, a gamer’s stream... There are more than 50 billion devices, cameras and sensors on the planet collecting a vast amount of information about us – personal and public, often without our consent.

Earlier, the humanity has not been able to process such huge amounts of data, but artificial intelligence can do it in a matter of minutes. And how this superpower will be used depends only on whose hands it falls into.

After Cambridge Analytica learned how to use Facebook users’ data to shape their political beliefs and influence the outcome of the election process, the UK and the EU have become innovators in the field of personal data and AI regulation. The European Parliament recently drafted the AI Act, the world's first legislation to regulate the use of AI.

Useful AI

“We have made history today,” Brando Benifei, a member of the European Parliament working on the EU’s artificial intelligence law, told reporters.

Brussels hopes that the law will start a trend of developing global standards for AI technologies, which have become a “superstar” in recent years and a driver of unprecedented change in a wide range of industries: from chatbots, such as OpenAI's ChatGPT, to surgical procedures, bank fraud detection and counter-terrorism.

The law “establishes harmonised rules in the field of artificial intelligence”, according to the AI Act’s explanatory memorandum. Parliamentarians recognise that the “family of technologies” using artificial intelligence can bring “a wide range of economic and social benefits across the spectrum of industries and social activities”.

These include more accurate forecasting, optimised resource allocation and personalised service delivery. The use of AI can also contribute to positive social and environmental outcomes and bring fundamental benefits to businesses and the European economy. This is especially true in the public sector and in the areas of environmental protection, healthcare, finance, mobility and agriculture.

Dangerous AI

However, the same miracle tool, says the draft law, can “create new risks or negative consequences for individuals and society as a whole”. Given the pace of technological change and the challenges it poses, the EU is seeking a balanced approach.

“It is in the Union interest to preserve the EU’s technological leadership and to ensure that Europeans can benefit from new technologies developed and functioning according to Union values, fundamental rights and principles.”

Therefore, the law should make artificial intelligence “human-centred and trustworthy” and prevent possible negative impacts of the technology on security, democracy, transparency, privacy and other inalienable rights.

To this end, it is proposed that AI systems be analysed and differentiated according to their level of risk. Those deemed unacceptable will be restricted or banned. The “black list” could include technological developments of so-called “social rating”. Such products were introduced on a massive scale by Communist China in 2014. Since then, digital social rating systems have been continuously monitoring the lives of the country's citizens. The information collected and processed by AI includes bank account status, expenses, entertainment, social media activity, criticism of the government, messaging with colleagues and family, and patterns of behaviour and mood.

The Members of the European Parliament also decided to classify remote biometric identification systems in public places as “dangerous”, namely:

• biometric systems to distinguish people by gender, race, ethnicity, nationality, religion, political views, etc;
• police “predictive” systems (based on profile, location or previous problems with the law);
• emotion recognition systems in law enforcement, border control, workplace and educational institutions;
• systems that can create databases of faces based on the processing of images from cameras in public places, etc.

It is worth noting that a few years ago, following a series of terrorist attacks, European countries experienced a boom in the development of total surveillance systems in public places. In particular, solutions that allowed AI, for example, to analyse the slightest changes in a person's behaviour or facial expressions in microseconds and make assumptions about how potentially dangerous their next actions might be.

The issue of “recognition systems” remains the most sensitive in the negotiations between the law's authors, the European Council and the European Commission. Brando Benifei, co-author of the parliamentary committee on artificial intelligence that pushed the bill through to the vote, said the facial recognition law would provide “a clear guarantee to avoid any risk of mass surveillance”.

He stressed that the technology could be vital in the fight against crime, counter-terrorism intelligence and the search for missing children. However, some EU leaders are expected to call for a complete ban on biometrics to avoid increasing the risk of abuse by the police.

The Members of the European Parliament also identified high-risk systems that could pose a significant threat to health, safety and fundamental rights. These include AI systems that could influence voters’ opinions and election results.

Legal AI

Developers will be required to register their product ‘models’ in the EU database before they can enter the market. AI-generating systems based on such models - such as ChatGPT and OpenAI – will have to meet transparency criteria and provide a detailed list of databases used to “train” artificial intelligence, including: publishing all works by scientists, musicians, illustrators, photographers and journalists used to train them; and developing safeguards to prevent artificial intellect from generating content that violates the law. In particular, the creation of deep fakes and copyright infringement.

Only research and development products and AI components provided under an open-source licence are exempt from this rule.

European Parliament President Roberta Metsola called the AI Act “legislation that will undoubtedly set the global standard for years to come”. She said the EU now has the ability to set the tone globally and that a “new era of governance” has arrived.

Meanwhile, on the other side of the Atlantic, the US is still debating the “threat of human extinction”, posed by AI technology. “While big tech companies are sounding the alarm about their own creations, Europe has come up with a concrete response to the risks that AI is beginning to pose,” says the bill’s author, Benifei.

AI raises many social, ethical and economic questions. But a ban is unlikely to provide answers. In times of fundamental change, it is important to act quickly and take responsibility and apparently, the EU likes to be a trendsetter in this regard.