It was a transatlantic Google search. In Brussels and Washington, we simultaneously searched “Artificial Intelligence and Copyright.” The US search returned an AI-powered summary. The European results listed ten blue links.

Google calls these AI-generated answer texts “AI Overviews.” The feature launched in the US in May 2024 and is being rolled out to one billion global users in over 100 countries.

But not in Europe.

Leading US tech companies–including Google, Apple, and Meta–have delayed launching new AI products and services in Europe. The delays underline a growing risk: will Europe’s tough digital regulations leave the continent as a tech laggard? Many European policymakers believe the companies are slowing rollouts to make a political point, undermining the new rules on purpose.

The companies insist that they are just being prudent. They cite several European regulations that make them hesitate, including the General Data Protection Regulation, the Digital Markets Act, and the AI Act.

Privacy rules represent a crucial barrier. Complying with the General Data Protection Regulation has not proved easy for Big Tech. Google postponed the release of its AI chatbot Bard last year after the Irish Data Protection Commission expressed concerns. Gemini–an updated version of Bard–faced delays in Europe, and our transatlantic search suggests its limited availability. YouTube’s latest AI features, powered by Gemini, are only available in the US.

The legal basis for collecting European data to train large language models remains unclear. AI developers need to gain permission from users – a tall task. In May 2024, Meta sent over two billion notifications to inform Europeans that their public posts on Facebook and Instagram would be used to train Meta AI provided they did not fill in an objection form.

But Meta was forced to pause when Ireland’s Data Protection Commission said the plan violated GDPR. In a blog post, Meta spokesperson Stefano called the decision “a step backwards for European innovation.” The owner of Facebook and Instagram decided not to launch its latest multimodal Llama model–capable of understanding text, audio, images and video–in Europe, citing the “unpredictable nature of the European regulatory environment.”  Meta now offers European customers its Ray-Ban Smart Glasses without AI features.

Both US and European companies are calling for regulatory certainty. In September, the Meta drafted an open letter signed by over 50 companies including European tech champions Ericsson, Klarna, and Spotify. Without access to European data, the letter says, AI models “won’t understand or reflect European knowledge, culture or languages.” EU policymakers, it reads, “face a choice between modernizing GDPR legislation” or “watch as the rest of the world builds on technologies that Europeans will not have access to.”

The Digital Markets Act adds further complications. When Apple launched its latest “Apple Intelligence” features, developed in collaboration with OpenAI, it didn’t bring them to Europe, blaming “regulatory uncertainties brought about by the Digital Markets Act.” The DMA lays out strict anti-trust rules, including interoperability requirements that mean Apple needs to ensure that users can choose other AI services and use Apple Intelligence on competing devices. Apple claims this brings unacceptable privacy and security risks to users and has asked for more clarity from the European Commission on device sharing and the level of access it needs to grant third parties.

The new AI Act could represent yet another barrier. It prohibits certain uses of AI, including social scoring, biometric categorization, and emotional recognition in workplaces and educational institutions. Uncertainty surrounds the legality of AI models capable of recognizing emotions from speech, such as OpenAI’s Advanced Voice Mode for ChatGPT. How the law will be interpreted remains to be seen.

Although European policymakers are scrambling to boost Europe’s slowing productivity rate and low competitiveness on AI, they remain adamant about the need for tough regulation. “Nobody says that Google and others cannot introduce new technologies in Europe,” says EU Commissioner Vera Jourová, who led work on the GDPR. “Maybe, one, two months, half a year later than somewhere else, but we want to be sure,” she says, that “innovations are developed to do good to people.”

She may be right. After the delays, companies are releasing products. Most will find the European market too lucrative to ignore and choose to bow to EU regulation. Google has agreed to enhance transparency and user protections on its AI models in Europe. English-language Mac users in the EU can now access Apple Intelligence and the company plans a European roll-out on iPhones and iPads in April. OpenAI launched its Advanced Voice Mode for ChatGPT in Europe.

Expect the skirmishing between Silicon Valley and European policymakers to continue. European consumers will suffer, at least in the short run. But in the end, they may end up with safer products.

Oona Lagercrantz is a researcher with the Digital Innovation Initiative at the Center for European Policy Analysis. She earned a first-class bachelor’s degree and a master’s degree with distinction from the University of Cambridge, specializing in the politics of emerging technologies.

Bandwidth is CEPA’s online journal dedicated to advancing transatlantic cooperation on tech policy. All opinions are those of the author and do not necessarily represent the position or views of the institutions they represent or the Center for European Policy Analysis.

The post Europe’s AI Blues: US Companies Slow Deployment appeared first on CEPA.