How (and why) to fix political ads in the EU
Meta follows Google (and TikTok and Microsoft) and leaves the EU political advertising market. The damage will be significant.
Sorry for the relative radio silence. We’re working on lots of new things, much more news on that to come. Follow us on BlueSky for more regular updates about election ad spending and activity. For ad spending in current elections, check Trends.
The post below is quite long, but we’ve tried to be as comprehensive as possible in our analysis of the wide-ranging impacts of Meta’s decision to join others in banning political, social and issue ads in the EU, as a response to forthcoming regulation.
The summary is:
Banning political ads favours those whose speech is already highly amplified by social media algorithms as well as bad actors.
There will be considerable secondary effects on charities and government communications.
It’s not possible to fully ban political ads, and no one has successfully done so.
There’s a certain hypocrisy to US companies “censoring” the speech of Europeans, as their own government rails against the EU “harming” the free speech of Americans.
Fixing the mess will require the EU finding a way to prevent platforms discriminating against political advertising in their commercial practices, likely by weakening existing legislation and changing the mechanism by which it is enforced.
If this happens, the benefits will be significant. It will create a fairer, more competitive political advertising market, wider reach for legitimate political speech, more transparency and accountability, and lower compliance costs for platforms.
What happened?
Last Friday, Meta announced it would ban “Political, Social and Issue” ads in the EU, starting in October. This has likely been coming for a while, but is a huge decision and will have a large effect on the way European politics is done online. Meta cited the forthcoming implementation of the EU’s Transparency and Targeting of Political Advertising (TTPA) regulation, focusing specifically on the targeting provisions of the new law and the perceived effects on their products.
Some will rejoice. Political ads can be annoying, particularly if a politician you don’t like is repeatedly thrust into your social media feeds. After the Trump and Brexit referendum of 2016, people incorrectly ascribed Facebook ads with magical powers of persuasiveness. Surely stopping them will also stop bad things happening again?
Unfortunately, that’s wrong. There are no winners here - not Meta, not the EU, not parties, campaigns or charities, not voters, not advocacy and research organisations like ourselves, not democracy. Below, we set out what it means, how we got here and what needs to happen next.
What does an ad ban mean for political speech?
The primary result of Meta’s ban is that online political speech in the EU will become even more algorithmically mediated. Political actors will have significantly less control over whether anyone sees their material than they do now.
For some, who built their politics and careers around online attention and engagement, the ban will further extend their advantage. By contrast, the relative losers will be more “normal” politicians, who prefer to talk about policy or don’t have any particular social media magnetism, as they can no longer pay for speech and reach on the largest platforms. As a consequence, we’ll see many more politicians flailing around, trying to generate engagement, “go viral”, pay “creators” or hire networks of accounts off-book to try and earn them their own tiny drop of attention.
Following from this, political speech in Europe will continue to get more extreme, as politicians search for whatever juice the algorithms can give them. “Reasonable” online politics will be harder to find. Not everyone can be a Zohran Mamdani (a charismatic, young, video-first politician running in a world-famous city with a giant media focus on him during an off-cycle primary election when not much else was going on). By contrast, a candidate for local elections in (say) Poland or Spain, competing with thousands of other candidates nationwide, only needs a few hundred zlotys/euros worth of ads to reach their specific pool of voters (rather than a full-time videographer, designer, copywriter and stylist, let alone actual charisma, good looks, unique policy positions etc).
A further disadvantage, that all politicians will face, is that it will be more difficult, perhaps impossible, for them to reach their voters. Paid online advertising allows them to show ads to people of voting age, in quite tightly defined geographical areas, often with specific interests or characteristics. Social media algorithms don’t work like that, creating a jumble of content ranked on the basis of predictions about what a user might be interested in or engage with, with the goal of keeping them on the platform to see (ironically) more advertising. An organic social media post by a local politician in Paris is of little use to a voter in Perpignan. The ability of a campaign to be intentional and efficient about who it reaches (or doesn’t reach) is therefore greatly reduced when it can’t buy ads. Millions of Europeans will now see “Vote for me today!” messages from politicians they couldn’t vote for a week after their election has happened, because the platform algorithm decided it was a good time to show them the post. This is bad for campaigns, voters, democracy and for Meta too, because their services will show less relevant political content.
Another set of losers from Meta’s ban, and hugely important, are European charities, who spend millions a year on ads to find new donors and supporters. People are starving in Gaza, yet Meta will cut off Medicins Sans Frontieres, UNICEF, the Red Cross and many others from fundraising for their work there in less than two months. Money for vital causes will be lost, with serious consequences. For this reason alone, Meta’s decision is immoral. Charities have until October to find the same money elsewhere. This will be impossible.
Government communications in the EU will also be affected. Many national and local governments, and their agencies, use Meta ads to promote things as various as the opening of a new park, to media literacy, to the rollout of social programmes and public health initiatives. Ads let them target the specific groups of citizens who will benefit from these. Soon, no more. “Make videos the algorithms will like” isn’t a realistic replacement for this for slow, often necessarily conservative government communications. Or do you think your government should post like the White House does?
The final group to mention is the ecosystem of thousands of staff, agencies and consultants who work in political and charitable advertising in the EU, who spend their time trying to get the message of their party or client out. The social media platforms are so large, and the skills and knowledge used to work within their services so specific and detailed, that they are economies in themselves. Many have built businesses (or parts of their businesses) around running online political advertising, of which Meta was the largest portion. They will need new jobs.
Why bans don’t work (and hurt even more people in the process)
One of the problems with Meta’s decision to ban ads is that the definition of “political” is hard to pin down. The company’s original decision, in 2018, to create a broad definition including “social and issue advertising” in their political ad programme was designed to help with this lack of precision. By casting a wider net, Meta was more likely to verify and make transparent a wider range of advertisers seeking to influence elections and policy.
However, as a direct result of this original ambition, their ban will now affect a much larger number of actors. In October, the majority of political advertising will disappear from Meta’s services, because it’s easy to identify ads from mainstream political parties and candidates. The company already has a large list of political and issue advertisers who have gone through their verification process to use for this.
But political, social and issue ads will continue to run around the edges, as new accounts pop up and run them accidentally (or on purpose), albeit now without the transparency, verification, moderation and oversight of a well-managed political advertising service.
The ban will therefore make it harder to detect bad actors who don’t follow the rules, and will ensure such accounts have a significantly higher share of voice than they do today. For example, ads from foreign influence campaigns, though likely still rare and low-reach, will now be “louder” than those of the CDU in Germany, Fratelli d’Italia, VVD, PSOE or any other European political party.
Worse, because of blurry definitions and inconsistencies in the way Meta’s systems work, their advertising systems create thousands of false positives each year - ads and advertisers that are not political, but which are identified to be. Things as prosaic as companies installing solar panels, or musicians and authors who, because they advertise under their own names often “look” like politicians, will be denied advertising. Their speech, marketing and potentially entire business will be accidentally caught - and destroyed - in the crossfire.
Our point is this - it isn’t possible to reliably ban political ads on large social media services that accept self-service advertising. There are too many ads and advertisers, too many bad actors and far too many shades of grey. Meta, like TikTok and Google before them, are choosing to fall into this trap. When it is inevitably proven that political ads are still running on their services, there could be investigations, legal cases and fines. After all, the TTPA will still apply to them, with all its attendant penalties. What mitigation will they now be able to plead when there are no good political ads, only bad ones?
Why are Meta doing this?
Let’s now turn to the regulatory “uncertainty” that led Meta to this decision.
When the TTPA was passed, alongside an attempt to define a political ad and transparency provisions, it included a ban on the “profiling” of the targets of political ads using special category data (as defined by the GDPR). Like a food label which tells us a product “may contain traces of nuts”, Meta’s is saying it can’t guarantee special category data isn’t included in the targeting mixture for any form of advertising. Because it can’t disentangle, let alone make transparent, such data in targeting, a ban is the only way forward.
We think the legislation over-stretched. What should have been a regulation to set common transparency standards and advertiser verification processes (both of which Meta are broadly fine on), helping organisations like ourselves, as well as journalists, researchers and other political parties hold campaigns to account, instead now includes provisions that made it harder to convince the largest tech platforms it was in their interest to comply. Limiting targeting methods in the TTPA made the definition of “political” vastly more important because it meant Meta had to offer different services for different types of ads. To do this, a service needs a way to draw a clean, straight line between them but, as explained above, this simply doesn’t exist.
That said, Meta has been slippery and cowardly. Nick Clegg, when he was still at the company, argued Meta wanted and welcomed the clarity regulation would bring. Well, the TTPA is the EU’s regulation for political ads, and they clearly don’t welcome it. Added to this, while Google made their equivalent announcement last year, Meta waited until late on a Friday afternoon when Brussels is largely on holiday, 75 days before the regulation was due to go into full effect. They waited and waited, and let things drift on. It’s true that the EU Commission has taken a very long time to provide the details and specifications needed to complete the TTPA. Too much has been left to secondary legislation and even now Brussels hasn’t finished all the details. But as far as we know, Meta hasn’t raised this as an issue. There was no “final straw”. The waiting seems to have been a tactical ploy in the grand game of the EU vs the US (Meta also recently declined to sign up to a voluntary code of conduct on AI).
The unintended consequence of the TTPA is not that it weakens Meta (or Google) by forcing them to ban political ads (as they say again and again, there’s not much money in it for them, it’s no great material loss), but the regulation will (unintentionally) weaken European democracy. It was well-meaning, but in its current form, by including limits on ad targeting, it’s proved to be an own goal.
Two notes on the wider context
It’s worth noting at this point that TTPA enforcement will be quite different to that of the DSA, with more emphasis on the role of the EU states and by different entities (e.g. electoral management bodies rather than media regulators). Given contemporary geopolitics, where the US treats EU regulation of “their” tech platforms as a bargaining chip in a wider trade war, we think it’s currently extremely unlikely that a EU member state would take action against a platform for failing to meet their obligations under the TTPA. Perhaps Meta (and Google’s) lawyers perceive a technical risk, but it’s certainly not a practical one at this stage. The idea that a US platform would be forced to pay a giant fine for failings in political ad transparency systems is for the birds. Furthermore, the TTPA also contains a review clause, which could be used to amend it (we think) as soon as 2026 (the law was passed in March 2024 and the review is “within two years”). Being cynical, we’d expect no cases to be brought in that timeframe, and for the platforms to start dictating the terms of a revised regulation or for it to be scrapped entirely.
The second note is an irony. Representative Jim Jordan’s Judiciary Committee of the US House of Representatives describes the DSA as “the EU’s comprehensive censorship law”. And yet here are giant American platforms denying Europeans the promotion of perfectly legitimate political speech that might reach hundreds of millions of citizens. The impact of this is far broader than any “European DSA rules allowed some tweets by my ideological friends to be taken down” complaint. A further irony is that a large amount of American speech (aka money that supports European campaigns of all stripes) will also be curtailed as a result of these bans. Is anyone going to support their free speech rights?
American platforms will censor European political speech from the centre, left and right (including many of Jordan’s ideological fellow travellers) more directly and at a larger scale than anything the EU has ever imagined. Yet we don’t expect any outcry from Washington.
Will anyone actually benefit from ads disappearing?
Hungary’s next parliamentary elections are in April 2026. For years, Viktor Orban and Fidesz have dominated the Hungarian media landscape, including of paid political advertising. When we speak to Hungarian civil society and opposition, they sometimes argue that a ban would help their cause, as it negates one of the many advantages Orban holds. But, unfortunately, an oligarchic semi-democracy is an oligarchic semi-democracy. Algorithms which already favour extreme content will noisily promote Orban’s anti-immigration, conservative message, and will be supported by an opaque army of paid influencers and media outlets. What happened in Romania last year may happen again in Hungary, though this time there will be no constitutional court to annul the result, and the opposition will be less able to use paid speech to counter it.
Others argue that voters will benefit, because political ads are invasive and annoying. They also sometimes contain mis- or disinformation or misuse personal data (though the vast majority don’t). It’s our view that a mixture of user controls (most platforms allow you to block advertisers or categories of advertising) and transparency is a much better solution. We know that’s not perfect, nor will it ever be, but in our view it’s the best balance between the necessary trade-offs.
This leaves one final set of beneficiaries - the bad actors described throughout this piece, who already enjoy algorithmic amplification, who are happy to skirt electoral rules and who challenge the very existence of democracy. If the only people to truly benefit from this are the ones we’d all least like to, we’re in a very bad place.
The fix: All platforms need to take political ads seriously
Fortunately, the policy solution is simple (though the politics will be complicated).
The EU needs to find a way to prevent platform discrimination against political advertising. If a very large platform accepts ads (they all do), they really should accept political (and social and issue) ads. There’s no obvious legal mechanism to make this happen, but one way would be to say “if you ‘ban’ political ads, and we find any, the penalties could be large. But if you allow them, and follow the law, we will be understanding of the many difficulties of doing this at scale.” In other words, we’ll tolerate a proportion of false negatives and positives provided you ensure most ads are verified and transparent.
The TTPA’s clause forcing a review of the legislation in 2026 is an opportunity in that respect. This could start soon, with a view to finding the right balance of carrot and stick. Done right, it could reduce the large platforms desire to duck the TTPA, making it less likely the would continue to slice off political advertising portion and sacrifice it to regulatory “complexity”.
This would be good in several ways. One is a very “why the European Union exists” benefit. More participants in the market, operating under common rules and standards, is why the EU exists. It would lead to a wider choice of services for political advertisers (Meta, Google/YouTube, TikTok, X, LinkedIn), the ability to reach a broader range of audiences (especially younger voters), as well as more innovation and product development in the market. Costs for campaigns would likely fall. Users would be safer.
The market would also be fairer for the platforms. Meta being the last major player standing in the EU meant it faced all of the regulatory scrutiny and cost alone. With more players in the market, that cost is spread and lessons can be learned and shared as new challenges and services emerge.
Lastly, democracy would benefit. A transparent political advertising market is what we’ve consistently called for since we started this work the best part of a decade ago. There should be common definitions for “political and issue advertising”, reliable advertiser verification and good quality ad libraries. There should be an understanding that there will always be false positives and negatives and that only egregious errors or flagrant disregard for the rules need to be punished. From this, there should be a thriving ecosystem of organisations holding political advertising to account, rather than having to spend time and money hunting for it in the darker corners of the internet.
Political advertising is unloved, but Churchill’s famous line that “democracy is the worst form of government, apart from all the others” applies here too. An internet without political ads, where algorithms are the primary deciders of what voters do and don’t see, where there’s little transparency, and where the good guys who follow rules are punished while the bad guys who don’t, is worse than an internet with them.
The EU needs to work, quickly, with the platforms and other stakeholders to find a way forward.