Germany urges tightening of draft European AI regs

By on 01/07/2020
Germany wants proposals for European AI regulation to be stronger. (Image by Deepak pal via flickr).

Rules governing artificial intelligence (AI) security at the EU level need to be strengthened to ensure ethical use of the technology, the German government has said.

Berlin was responding to the European Commission’s white paper on AI, published in February. While it said it broadly agrees with the proposals, the German government’s response sets out its concern at the idea that only AI applications designated as “high risk” should have to meet special requirements, according to European news website Euractiv.com.

‘High risk’ applications are defined by the commission as those that are used in sensitive sectors such as health, security or justice, and where their application is also associated with risks such as danger to life, injury or discrimination.

The commission wants these types of AI applications to have extra legal requirements, governing fields such as human oversight, data and record keeping, robustness and accuracy.

‘High risk’ classification

But Germany, which this week took over the revolving presidency of the commission, does not believe that these requirements are far-reaching enough. It has proposed tightening both the classification of ‘high risk’, and the requirements themselves.

The criteria for defining an application as ‘high risk’ should be reconsidered and extended if necessary, the German government said. It fears that limiting special requirements to particular sectors could mean that other high-risk uses would not be covered by the regulation.

Berlin believes that a new scheme should be developed containing several levels of classification, taking into account the amount and probability of damage ­– including harm to life and property, democratic processes, environment, climate, and societal and economic participation.

The German government also wants a clearer definition of when data records must be stored on a mandatory basis, and for how long. The commission has said this is necessary “in certain justified cases”, with the requirement of storage for “a limited, reasonable time period”.

Finally, Germany has asked for more detailed definition on the circumstances that would mandate human supervision. Brussels has yet to set these out.

Future proof laws

The commission wants to pass a first law for AI early next year – aiming to challenge the dominance of the US and China in the emerging technology, while retaining a strict focus on ethics in order to protect EU citizens.

A group of over 120 researchers, including some of the world’s leading AI scientists, last week applauded the commission for its efforts to determine the role that government should play, and for supporting regulation of AI systems in high-risk applications.

An open letter from the group warned the commission to stand strong against any attempts by industry to water down the rules. It also noted that the rules must be capable of handling or adapting to the many forms and capabilities that AI could develop in the future.

“We now sit in the early days of AI, and the choices we make over the next decade will crucially shape its place in and relation to society,” the letter says. “The stakes are high, and the potential ability of AI to remake institutions means that it is wise to consider novel approaches to governance and regulation, rather than assuming that existing structures will suffice.”

About Catherine Early

Catherine is a freelance journalist and editor specialising in government policy and regulation. She writes predominantly about environmental issues and has held permanent roles at the Environmentalist (now known as Transform), the ENDS Report, Planning magazine and Windpower Monthly, and has also written for the Guardian, the Ecologist and China Dialogue. She was a finalist in the Guardian’s International Development Journalism competition 2009, and was part of the team that won PPA Business Magazine of the Year 2011 for Windpower Monthly. She also won an outstanding content award at Haymarket Media Group’s employee awards for data-led stories in Planning magazine. She holds a 2:1 honours degree in English language and literature from Birmingham University.

Leave a Reply

Your email address will not be published. Required fields are marked *