I have spent the better part of three decades being paid to look closely at gambling businesses. Due diligence, it is called, and the pitch to clients is always the same: give me the data room, a meeting with management, a review of the product offer and the competition, give me a few weeks, and I will tell you what you are actually buying, as opposed to what the PowerPoint says you are buying. It has proved to be a marketable skill. It has also, this month, produced a useful case study, courtesy of the Court of Justice of the European Union and a company that looked closely enough to sign the deal, then argued in court that it hadn’t really been looking at all.
On 16 July, the Court handed down judgment in Case C‑421/24, AGCOM v Google Ireland, and if you advise anyone whose business model involves hosting, monetising or vaguely tolerating other people’s content, it is worth five minutes of your time.
The facts are unremarkable in the way that most important cases are unremarkable. Italy’s communications regulator, AGCOM, fined Google Ireland €750,000 in 2022 and ordered the removal of 630 YouTube videos promoting online gambling, in breach of Italy’s Dignity Decree, which bans gambling advertising in more or less any form you can invent. The videos sat on channels enrolled in the YouTube Partner Programme, where Google, YouTube’s owner, shares advertising revenue with creators who clear a subscriber threshold. Google argued it was merely a host: a passive shelf on which other people’s videos happened to sit, protected by the safe harbour that EU law grants to neutral intermediaries. An Italian court agreed with it but AGCOM did not, and the case eventually landed in Luxembourg.
The Court split the question in two, and the first half went Google’s way. AGCOM had argued that because gambling itself sits outside EU e-commerce law, on account of the “profound moral, religious and cultural differences” between member states on the subject, hosting a gambling advertisement should sit outside it too. The Court disagreed. Storage is storage, it said, whether the thing being stored is a betting advert or a video of a cat failing to land a jump. So hosting stays squarely inside the ordinary rules.
That first ruling got a fraction of the attention the second one did, which is a shame, because it was the more important of the two. AGCOM’s underlying argument was not really about Google’s revenue-sharing habits. It was an attempt to knock gambling advertising out of the safe harbour outright, for every platform, on the basis that hosting it is as morally loaded as hosting the bet itself. Had that argument won, no ad network, exchange or affiliate pipe carrying a gambling advertisement into Italy would have had access to Article 14, however passive its role. Neutrality would have stopped being the test; presence would have been enough.
The Court declined to go that far, and the practical effect is that the machinery which carries most illegal gambling advertising into Italy, programmatic exchanges, affiliate networks, algorithmic ad placement, keeps its safe harbour intact, provided it stays genuinely passive. Google lost because it did something active: it vetted a channel and then took a cut of the revenue. Most of the infrastructure moving adverts for the unlicensed operators behind Italy’s estimated €20 billion illegal market does neither. AGCOM won its case against one platform’s partner programme. It did not win the tool it most likely coveted, which was the power to hold the wider advertising ecosystem liable for merely carrying the content, without first having to prove knowledge or control.
None of which helped Google itself, because the second question turned on its own conduct rather than the wider rule. To keep the hosting exemption, a platform has to behave as a strictly technical, automated, passive intermediary, with no knowledge of and no control over what it stores. Before letting the creator into its revenue-sharing programme, however, Google had reviewed the channel: its main theme, its most popular and newest videos, the metadata. Not every video, mind you. Just enough to decide whether the channel was worth doing business with.
That, the Court ruled, was enough to sink the neutrality claim. Knowledge of a channel’s “essential content” defeats the exemption just as surely as active control over it does; the two tests are independent, and you only need to fail one. Google had not stumbled on the gambling content by accident, nor had a third party tipped it off. It had gone looking, for sensible commercial reasons, and the looking is what led it to stumble.
There is something almost too neat about this, and I say that as someone who a large part of their professional existence rests on the premise that looking is virtuous. The Margaret Heffernan thesis in Wilful Blindness is that organisations get into trouble by choosing not to see things they would rather not know. The AGCOM judgment does not overturn that. It simply adds a second, narrower trap next to it: an organisation that looks, finds something, and does business anyway cannot claim afterwards that it hadn’t seen. Both are failures but only one of them used to come with a safe harbour attached.
European Court Judges have form on the second trap. A month before AGCOM, in the joined WebGroup Czech Republic and Coyote System cases, the Grand Chamber found that an operator whose algorithm decides how content gets ranked and distributed has exercised “control” and loses hosting protection, even if no human at the company ever watched a frame of it. Between the two rulings, the space in which a platform can plausibly claim to be a mere pipe, rather than an active participant, has narrowed. Vetting a creator, curating a feed, running a brand-safety check, all of it now risks being read back to you in court as evidence that you knew.
“Wilful blindness has never been much of a legal strategy, and it is a worse business strategy.”
For anyone advising operators, affiliates or platforms on the Italian market, or frankly on any market with an advertising regime worth the name, the practical lesson is uncomfortable, but it is not a licence to see less. The due diligence you run to reassure yourself, or your advertisers, that a channel or a partner is safe is the same due diligence a regulator can later use to prove you knew. That should trouble you rather less than the alternative it seems to invite, which is not looking at all. Wilful blindness has never been much of a legal strategy, and it is a worse business strategy, whatever the CJEU has just done to the economics of it. My advice to clients has not changed: look properly, and act on what you find. What has changed is that the law now has a sharper way of catching those who do the first and then skip the second.
There is also a paradox sitting underneath the Italian facts. The Dignity Decree’s blanket advertising ban has done what blanket bans usually do (think Prohibition in the USA): pushed promotion into exactly the channels regulators struggle to police, while leaving the compliant end of the market to fight over the fine print of hosting liability. The less comfortable question for regulators everywhere, and not just AGCOM, is whether a rule that makes ignorance the safer legal position is a rule anyone should want. Good vetting ought to be rewarded, not punished for leaving a paper trail.
For now, the more immediate question is for every platform, affiliate network and marketing partner working out how to keep vetting their partners properly while making sure that vetting actually leads somewhere, rather than sitting in a file as evidence of a concern nobody acted on. That, and not looking away, is the only version of this I am prepared to recommend to a client, however much simpler the CJEU has just made the alternative sound.