Brussels Made YouTube Explain Its Algorithm. YouTube Made Sure Nobody Could Find the Explanation.
- Michael Routhier
- Jul 28
- 5 min read
Every time you open YouTube, an algorithm you've never seen makes a decision about you in less than a second, deciding what you'll watch next, how long you'll stay, and how much of your attention gets sold to advertisers before you even scroll. For years, that algorithm operated as a complete black box. Then the European Union stepped in with real, binding law and forced YouTube's hand. So why, three years later, can almost nobody in the EU actually find or use the transparent alternative they were legally promised?
That question is the entire story, and once you understand the answer, I think you'll be just as frustrated as I am.
The Law That Was Supposed to Fix This
Back in 2022, the EU passed the Digital Services Act, landmark legislation designed to end the era of platforms setting their own rules in total secrecy. Two provisions matter most here. Article 27 requires platforms to explain, in plain and intelligible language, the main parameters driving their recommendations and why certain content gets suggested over other content. Article 38 goes further for the largest platforms, called Very Large Online Platforms, requiring them to offer users a genuine alternative feed that isn't built on profiling their personal data at all.
This wasn't a suggestion. YouTube was formally designated a Very Large Online Platform by the European Commission in April 2023, meaning these obligations are legally binding, not corporate goodwill. On paper, this looked like exactly the kind of algorithmic transparency digital rights advocates had been demanding for a decade.
What Actually Happened When Regulators Checked
Here's where this story turns from a policy win into a genuine scandal. In March 2026, the digital rights organization European Digital Rights, known as EDRi, filed a formal complaint against Google with Belgium's telecommunications regulator, backed by detailed research into exactly how YouTube implemented these legally required features. What they found should concern anyone who thought this problem was already solved.
To access YouTube's non-profiling recommendation option, the one required by law, users on the web interface must navigate through five separate menu levels. On the Android app, it takes seven. Nowhere in that entire process does YouTube's interface ever mention that a non-profiling alternative exists at all. Users can only stumble into it by choosing to turn off their YouTube Watch History entirely, an unrelated feature bundled together specifically, EDRi argues, to make the legally required option as unappealing as possible.
It gets worse. Once a user does find their way through that maze and disables profiling, YouTube's landing page goes completely blank, showing no recommendations whatsoever, just a large, permanent, impossible-to-miss button nudging them to turn profiling back on. Meanwhile, reactivating the profiling-based algorithm Google actually wants you using takes exactly two clicks. EDRi's complaint calls this exactly what it looks like; a "harmful design pattern" specifically engineered to comply with the letter of EU law while defeating its entire purpose.
This Isn't Just a YouTube Problem, and That's the Point
If you think Google is uniquely guilty here, a nearly identical case just played out in Amsterdam. A Dutch court ruled against Meta in 2025 after finding that Facebook and Instagram's interface designs made choosing a non-algorithmic feed so difficult that it constituted a "serious infringement" of user autonomy, the exact same legal theory EDRi is now applying to YouTube. A separate 2025 investigation by the Dutch digital rights group Bits of Freedom found comparable manipulative design patterns across TikTok, Snapchat, and other major platforms specifically related to recommender system transparency.
The pattern here isn't one company being clever. It's an entire industry treating "legally required transparency" as a compliance checkbox to be technically satisfied and practically buried, rather than a genuine obligation to inform users.
What the DSA Requires | What YouTube Actually Built |
Plain-language explanation of recommendation parameters (Art. 27) | Vague, generalized language buried in Terms and Conditions |
Directly and easily accessible alternative feed (Art. 27(3)) | Five to seven menu levels deep, unmarked and unlabeled |
A standalone non-profiling recommender option (Art. 38) | Bundled with unrelated Watch History settings, forcing users to give up both |
No repeated nudging back to defaults (Art. 25(3)) | Permanent, prominent button pushing users back to profiling-based feed |
Why This Matters Even If You Never Read the Fine Print
I want to be clear about why this is worth your outrage, not just your attention. The entire premise of the Digital Services Act was that informed users, given a real choice, could meaningfully push back against algorithms designed purely to maximize watch time and ad revenue rather than user wellbeing. If that choice technically exists on paper but is functionally invisible in practice, the law hasn't actually changed anything for the roughly 400 million YouTube users across the EU. It's just given regulators a talking point and given Google a legal shield to point to when critics ask why the algorithm still feels like a black box.
EDRi's complaint lays out something Google itself would struggle to argue against; the company has a direct financial incentive to keep this option hidden, since more time spent on the profiling-based algorithmic feed means more ads served and more revenue generated, a plain conflict of interest the current enforcement structure has done almost nothing to resolve.
What the EU Needs to Do Now
Passing a good law was step one. Step two, actual enforcement with consequences, hasn't happened yet in any meaningful way. Here's what I think needs to happen next, and it's not complicated:
The European Commission needs to act on EDRi's complaint under Article 65(2) DSA and formally investigate, rather than letting it sit in a regulator's inbox
Genuine accessibility standards need to be codified, not left to each platform's discretion, so "directly and easily accessible" can't mean five buried menus
Fines under the DSA can reach 6 percent of a company's global annual revenue, and it's time regulators actually use that leverage rather than treating these as warning-letter matters
Independent, recurring audits of recommender system accessibility should be mandatory, not dependent on nonprofit organizations like EDRi doing the investigative work regulators should be funded to do themselves
The EU proved something important here; it has the legal authority and the political will to demand transparency from the world's largest platforms. What it hasn't proven yet is that it's willing to enforce that demand once a company like Google gets creative about technically complying while practically ignoring the spirit of the law entirely.
Michael Routhier is the founder of Tech 4 Grown-Ups, providing honest, unfiltered digital literacy for adults 55+, and host of The Virtuous Machine, exploring the ethics and human cost of AI. Read by tech-curious readers in 50+ countries. Explore more at tech4grownups.com.
Comments