Europe is full steam ahead on tech policy

Home / Clinical Practice / Europe is full steam ahead on tech policy

Europe is full steam ahead on tech policy

illustration of a neon colored grid with images of a gavel, the EU flag and binary code

Illustration: Tiffany Herring/AxiosExit Content Preview

Europe has long been the first mover on significant tech policy, passing far-reaching laws in areas where Congress hasn’t advanced past the hearing or markup stage.

What’s happening: With the hype around AI at a fever pitch, Europe is moving ahead with rules as it also implements the next phases of its landmark digital content and competition laws, the Digital Services Act and the Digital Markets Act.

Why it matters: In the U.S., tech companies have avoided major new legislation regarding content or the way they have to treat competitors, and are playing an active role in legislative frameworks for AI.

But in Europe, the time to weigh in on proposals is close to being over, and compliance is a real, incoming hurdle.

How companies will adapt to EU rules will show the rest of the world what they’re capable of, even as they warn of the dangers of onerous laws.

The time and expense that will be needed to comply are likely to make companies less open to strict or overlapping rules from the U.S.

Details: The EU AI Act, first introduced in April 2021, is in a part of the EU legislative process called “trilogue,” a negotiation among members of the Parliament, Council and Commission.

Major issues of disagreement that remain include how generative AI should be regulated, the definition of high-risk AI, AI use for biometric surveillance in public, and implementation and enforcement, per Dragoş Tudorache, a member of the European Parliament, speaking at an event hosted by Stanford University last month.

The Digital Markets Act, which governs competitive behavior and requires companies to change how they treat industry rivals, will apply “gatekeeper” status to Alphabet, Amazon, Apple, ByteDance, Meta, Microsoft and Samsung because of market size and number of users.

Companies have a chance to respond to those designations now. TikTok has said it meets the criteria, but disputes its inclusion on the list.

If they are deemed to be official gatekeepers, they’ll know by Sept. 6. From then, they’ve got six months to comply with the DMA.

“The commission is currently consulting with the public on the compliance report templates and the description of consumer profiling techniques, two key pieces of gatekeepers’ compliance requirement,” Daniel Zhang, senior manager for policy initiatives at Stanford University’s Institute for Human-Centered AI, told Axios.

The Digital Services Act, which governs algorithms, transparency, illegal content and access for researchers, designated its Very Large Online Platforms (VLOPs) and Very Large Online Search Engines (VLOSEs) in April.

They are Alibaba; Amazon Store; Apple App Store; Booking.com; Facebook; Google Play, Maps, Shopping and Search; Instagram; LinkedIn; Pinterest; Snapchat; TikTok; X (formerly Twitter); Wikipedia; YouTube; Zalando; and Bing.

Zoom in: Amazon, for one, is fighting its DSA designation: “The DSA was designed to address systemic risks posed by very large companies with advertising as their primary revenue and that distribute speech and information,” a spokesperson told Axios in a statement.

“If the VLOP designation were to be applied to Amazon and not to other large retailers across the EU, Amazon would be unfairly singled out and forced to meet onerous administrative obligations that don’t benefit EU consumers.”

Some U.S. lawmakers think the government should be doing more to fight back against “discriminatory” EU measures like the DMA and DSA, per a June letter to President Biden from dozens of members of Congress.

“These policies, which almost exclusively target American companies, collectively impose sweeping new regulations, forced technology transfers, compliance costs, hefty taxes, and potentially significant penalties while giving homegrown European rivals a leg up,” they write.

The intrigue: In June, Axios’ Ryan Heath reported that 10 leading AI models including Hugging Face and ChatGPT scored poorly with responsible AI standards laid out in the Act, per research from Stanford.

“While the parliament’s draft of the [AI Act] is still being debated at the trilogue, what the study shows is that compliance is feasible. With sufficient incentives, companies can change their conduct,” Zhang said.

Leave a Reply

Your email address will not be published.