The latest shot in the escalating battle over how to regulate facial recognition has been fired, and this time it comes from Strasbourg.
Proposals due to be unveiled on Thursday by the Council of Europe call for tough curbs on how the new technology’s awesome and troubling power should be wielded by both companies and public authorities.
The guidelines, seen by the Financial Times, are not binding, but they carry weight. The 47-country council is the guardian of human rights in Europe and decades ago helped set a lead on evolving international rules on data protection.
But the world has changed since, and with China forging ahead on facial recognition, the EU faces the urgent task of establishing its own rules and using them to influence the global approach.
The Council of Europe proposals, drawn up by an expert committee consisting of representatives from member states, call for stringent safeguards against abuses of the fast-developing technology. The committee says facial recognition must not be used to hire and fire workers based on their moods or how engaged they are in the workplace. It also requires companies to obtain consent from workers to use facial recognition data.
The ideas come as officials in Brussels draw up their own definitions of what is acceptable. The European Commission last year tabled a set of tough guidelines as the EU outlined ethical principles that regulate the technology for its citizens.
Those proposals were criticised for being too bureaucratic, however, and were watered down with haste.
Brussels had initially toyed with the idea of banning facial recognition technologies in public places for five years, before executing a U-turn under pressure from member states.
A big part of the problem, according to the Council of Europe, is that facial recognition often gets it wrong. The guidelines call for “periodic renewal of data” to “train and improve the algorithm used”. They add:
“Each algorithm has a percentage of recognition reliability, both during its development and use. It therefore seems important to date and record this percentage to monitor its evolution.
Should its reliability deteriorate, it will be necessary to renew the training photos and therefore ask more recent photos to be provided. This will also enable to protect from the consequences of changes in the shape of faces.”
Reliability percentage records could be made available to individuals or companies, the proposals suggest, to enable the use of highly reliable technology.
“The highest possible level of reliability should be ensured, considering that the use of a facial recognition system might result in very significant adverse consequences for the individual,” they say.
For its part, the commission is also worried about the lack of transparency about how algorithms work and their tendency for bias. For example, if bias is detected in an algorithm used by an insurance company and the firm is found to make discriminatory decisions based on the technology, this would be deemed as breaking the rules.
The EU proposals are being debated by MEPs, who will have to pass the law. Some are demanding far-reaching reforms.
An amendment tabled by Emmanuel Maurel of the French Socialist party and leftwing French MEP Manon Aubry has called on the commission to consider a total ban on the use of the technology by law enforcement agents.
The amendment didn’t get strong backing from other legislators, but it underscored how deep concerns run over facial recognition.
“The debate on this intrusive technology that has the potential to infringe people’s fundamental freedoms is key,” Ms Aubry told the Brussels Briefing. “At the European parliament we will fight for a full ban because we are concerned about the damage it can cause to society.”
Chart du jour: hunting for unicorns
Stocks in Goldman Sachs’ non-profitable tech index surged in value last year. Writing in the FT, James Bianco thinks the climb is the result of a wave of money coming from retail investors trying to find the next Uber or Amazon. For more, the FT has launched a Runaway Markets series on what is driving investors during the pandemic.
Around Europe
-
The EU has escalated its public spat with AstraZeneca — and also risks a new fight with post-Brexit Britain — after demanding that the pharma giant divert vaccine supplies from its UK facilities to the EU. Brussels has also insisted that AstraZeneca agree to publish its contract with the bloc. After a meeting with AstraZeneca chief Pascal Soriot on Wednesday night, the EU asked for a “clear plan” about when the promised jabs will be delivered. (FT) Writing in The Guardian, Bruegel’s Guntram Wolff blames the EU’s slow vaccine rollout on the commissioner ordering “too few vaccines too late”.
-
The collapse of Italy’s governing coalition this week has complicated Rome’s bid to access nearly €200bn of EU recovery funds. Italy, as with all member states, has until April to finalise a national recovery and resilience plan and submit it to the commission, but the latest political turbulence is not helping matters. (FT)
-
Six human rights groups have launched a legal action against the French police for “systemic discriminatory practices”. It comes months after officers in Paris were caught on camera violently beating a black man. Morale within the French police is low, according to a survey, with many officers reporting a sense of “all against the police”. (NYT/Le Monde)
Coming up today
Join the FT’s Michael Peel in conversation with former UK commissioner Julian King, hosted by the EU-UK forum from 11am (CET).
javier.espinoza@ft.com; @JavierespFT
michael.peel@ft.com; @Mikepeeljourno
david.hindley@ft.com