The European Consumer Organisation (BEUC) is urging EU authorities to investigate generative AI’s risks now, before consumers suffer harm. The group says safety, data and consumer protection authorities should use the laws already in place to scrutinize AI systems and enforce those laws.
Why BEUC wants investigations to start now
BEUC says tools such as ChatGPT create new possibilities for consumers, but also raise concerns about how people may be deceived, manipulated or harmed. In a statement, deputy director general Ursula Pachl pointed to disinformation, discrimination amplified by existing biases, and fraud as potential consequences.
The call is aimed at authorities responsible for safety, data and consumer protection. Pachl said they should not wait for consumer harm to happen before taking action, adding that existing laws apply to products and services whether they use AI or not.
BEUC represents consumer organizations in 13 countries in the EU. It made the appeal alongside a report by its member Forbrukerrådet in Norway, titled “Ghost in the Machine: Addressing the consumer harms of generative AI.” The report argues that generative AI poses consumer harms and identifies several areas that warrant scrutiny.
Risks that can be hard for consumers to see
One concern in the report is that some AI developers, including Big Tech companies, have closed their systems to external scrutiny. That can make it difficult to understand how a system collects data or how its algorithms work. Without that visibility, consumers and outside reviewers may struggle to assess how a tool reaches its results.
Generative AI can also produce incorrect information alongside correct information, without making the distinction clear to users. That leaves people to judge whether an answer is reliable, even when they may not have enough information to do so.
The report also raises concerns about systems designed to mislead or manipulate, and about bias linked to the information used to build an AI model. Security is another issue: AI could be weaponized to scam people or breach systems.
These concerns connect AI governance to everyday consumer protections. The issues described are not limited to how advanced a system is; they include whether people can tell what they are interacting with, rely on its output, and avoid being targeted through it.
The EU’s approach to AI risk
EU discussions about AI risk predate ChatGPT’s rise in public awareness. The region began debating risk in 2020, initially framing the work as a way to build trust in the technology. By 2021, the discussion had shifted toward “high risk” AI applications, and some 300 organizations joined efforts to advocate banning some forms of AI entirely.
As the EU develops region-wide rules, the discussion has become more pointed. The source article reports that the EU’s competition chief, Margrethe Vestager, recently highlighted the risk of bias when AI is used in critical areas such as financial services, including mortgages and other loan applications.
The EU had approved its official AI law, which provisionally sorts AI applications into categories including unacceptable, high and limited risk. The law covers a wide range of parameters for determining which category applies. The next step described in the article is for the EU to work with individual countries in the bloc on the law’s final form, including which systems and actors fit within each category.
Oversight will shape what the rules mean
The EU said it wanted to complete that process by the end of the year. How readily countries agree will help determine the final categories and the scope of the law. The article describes the measure as the world’s first attempt to establish a legal framework for how AI is used commercially and noncommercially.
For BEUC, the legislation must protect consumers through public scrutiny and clear expectations for AI output. Pachl said generative AI systems should produce output that is safe, fair and transparent for consumers, and called for public authorities to reassert control over AI systems.
The group has previously spoken up during major regulatory debates, including early opposition to Google in long-running antitrust investigations. That history illustrates how advocacy can precede regulatory action by years. The debate over AI’s effects and the role of regulation may also take time, making the group’s call for immediate investigations a push to address risks while the broader rules are still being worked out.