EU keen to set global rules on artificial intelligence

After a draft white paper about the EU's position on AI regulation was leaked earlier this month, Google chief Sundar Pichai, on Monday (20 January), warned the bloc about imposing its own regulations and called for an "international alignment" on the core values of the future laws of the sector.

However, ahead of what is expected to be the fourth industrial revolution, the European Commission wants to ensure an "appropriate" ethical and legal framework for the development of AI, which promises to boost innovation while making EU citizens' rights a priority.

In November, commission chief Ursula von der Leyen pledged to develop AI legislation similar to the General Data Protection Regulation (GDPR), an EU law on privacy.

"It is not about damming up the flow of data, it is about making rules that define how to handle data responsibly," she told MEPs back then.

"For us, the protection of a person's digital identity is the overriding priority," she added.

According to Ursula Pachl, deputy director-general at the European Consumer Organisation (BEUC), an NGO in Brussels, "it is important that the European Commission has announced a legislative framework and is vocal about the ambition to become a global standard-setter in this area, much like it has done with the GDPR on data protection".

However, according to the commission's draft white paper - seen by EUobserver - this upcoming framework is likely to be built on existing legislation and risk-assessment or requirements for specific applications of these technologies rather than to plump for global sectoral requirements or bans.

Three 'promising' options

The leaked proposal suggests a few options that the commission is still considering for regulating the use of AI more generally - mainly to avoid breaches of fundamental rights, on safety issues, and on liability risks.

These include a voluntary labelling framework for developers (based on voluntary previsions for technology makers with binding conditions), imposing sectoral requirements for public administration (including the use of facial recognition), or mandatory risk-based requirement for "high-risk" applications (in sectors such as health care, policing, or transport).

The commission is also expected to amend existing EU safety and liability legislation to address the new risks.

The draft white paper emphasises the need to oversee the implementation and application of this technology to ensure compliance with EU law, although it says member states should decide whether to rely on existing authorities or create new bodies for the regulation of AI.

The text inidcates that mandatory requirement for "high-risk" applications, amendments to EU law and public enforcement are the most "promising" options to address the risks specific to AI.

However, some might say that while the EU wants to set the standards, over-regulation could limit Europe's capacity to innovate, especially in comparison to the US, where more "flexible frameworks" can be expected.

A ban? 'Not likely'

Additionally, the leaked proposal floats the idea of a three-to-five-year period in which the use of facial recognition technologies could be banned in public places to identify and address the possible risks of this technology.

However, "such a ban would be a far-reaching measure that might hamper the development and uptake of this technology," the commission writes, adding that "it would be preferable to focus on the full implementation of the provisions of the General Data Protection Regulation [GDPR]".

"The face recognition ban is mentioned in the paper, but the authors write twice that this approach is not recommended, [so] the European Commission might be 'considering' a ban, but it does not look like they will push for it," said Nicolas Kayser-Bril from NGO AlgorithmWatch.

The draft proposal highlights that such a ban will also necessarily foresee some exceptions.

Meanwhile, this regulatory safeguard could face opposition from many member states that have deployed these technologies, including Germany, France or Hungary.

According to Diego Naranjo, head of policy at NGO European Digital Rights (EDRi), "member states should stop deploying [facial recognition] technologies until specific EU legislation is adopted.

"There is no innovation in deploying mass surveillance of citizens. China is already doing it very effectively, while the US has developed the largest surveillance system worldwide as exposed by Snowden in his revelations," he added.

The European Commission is expected to unveil the white paper on the EU's apporach towards AI in mid-February and some of its policy recommendations are likely to change in the meantime.