It is a well-known fact that after just a few minutes on TikTok, you can find yourself deep diving down a personalised rabbit hole, whether that’s dance trends, political debates, or more worryingly, dangerous stunts and misleading information.
Now imagine this – you’re completing your nighttime routine — scrolling on TikTok until 2 am like every other student – when a new challenge pops up on your For You Page (FYP). At first, it’s just a clip of teenagers balancing on top of a moving car. The next night, it’s rooftop climbs, motorway dares and subway surfing. You see it again and again. Soon, it feels less like a coincidence and more like fate – so you try it yourself. Hours later, it’s front-page news, and you’re in trouble. Who’s to blame? You, the person who watched the video, or the algorithm that kept pushing?
This is no longer a hypothetical; courts in both the US and Europe are asking whether social media recommendation systems—the very feeds that keep people glued to their screens—can be held legally responsible when things go wrong. TikTok’s FYP, Instagram Reels, YouTube’s ‘Up Next’ – all of these are powered by algorithms trained to predict what will hold your attention. They don’t just show you what you follow; they shape what you see next. The question is simple: if an algorithm makes harmful content more likely to reach you, is that negligence on behalf of the company?
In legal terms, negligence requires proving a duty of care, a breach of that duty, and harm caused as a result. Plaintiffs argue that platforms are aware that their recommendation engines can cause damage and design them in a manner that makes that harm foreseeable.
The US Legal System and TikTok
For decades, American tech companies have been protected by Section 230 of the Communications Decency Act, a law that says platforms cannot be held liable for what users post. Courts often extended this to algorithms, treating recommendations as just another form of hosting. That changed with Anderson v. TikTok (2024). After a ten-year-old girl died attempting the ‘blackout challenge’ she saw on her feed – a viral dare where children choke themselves until they lose consciousness – her mother sued TikTok. The plaintiffs argued TikTok’s recommendation algorithm pushed the dangerous videos directly into children’s feeds, making it not just a passive distributor but an active promoter. The court stated that when harm arises from a platform’s own design choices – in this case, an algorithm designed to feed videos, which alters the speech from ‘user speech’ into TikTok’s own first-party speech. Once that happens, the Section 230 shield is gone.
However, the court stressed this doesn’t mean TikTok is automatically liable. It means TikTok can’t hide behind Section 230 and must defend itself on merit. Examining the Supreme Court’s 2024 decision in Moody v. NetChoice, which addressed whether platforms’ content moderation practices are protected under the First Amendment. The Third Circuit treated TikTok’s feed as TikTok’s own expression, and therefore outside the protection of Section 230.
Yet individuals cannot just walk into court and say, ‘The algorithm made me do it’ as a personal defence. But it does suggest that platforms could be treated less like a tool and more like designers of potentially dangerous products.
EU Law and Data Protection
The EU has taken a more direct approach by passing the Digital Services Act (DSA) and the AI Act, which force large platforms to assess the risks of their algorithms, limit harmful effects and submit to independent audits.
In February 2025, Europe launched its first major test of these acts, as TikTok and X (formerly Twitter) were hit with four class actions in Germany led by the Dutch Foundation for Market Information Research (SOMI). They now face charges of using manipulative algorithms and exploiting user data. Both platforms are accused of breaching the holy trinity of Europe’s digital regulation – the DSA, the General Data Protection Regulation (GDPR) and the AI Act.
TikTok is facing allegations for deliberately manipulating young users with addictive design strategies, exploiting children’s psychology to maximise screen time, and misusing sensitive personal data to fuel its recommendation algorithms. According to the suits, this violates the AI Act’s new ban on manipulative, deceptive, and exploitative AI systems. TikTok also allegedly spied on its users.
X is accused of repeatedly covering up major data breaches, failing to notify affected users, and processing sensitive personal data without a legal basis. Plaintiffs say its algorithmic profiling undermines freedom of opinion formation and accelerates divisions in public discussions.
Both platforms are charged with spreading misleading information, deepfakes, and undisclosed political advertising (especially around elections), which are exactly the abuses the DSA was designed to stop. At stake are billions in damages (€500–2,000 per TikTok user; €750–1,000 per X user).
Where does Ireland stand?
Ireland is in the spotlight too. The Irish Council for Civil Liberties (ICCL) has launched the country’s first-ever collective redress action against Microsoft, targeting its ad-tech system (Microsoft’s online advertising system). Every time you open a site or app, hidden code auctions off your attention to advertisers in milliseconds. To do that, ad-tech companies hoover up vast amounts of personal data – location, browsing history, even inferred details about your health or mood – and use it to decide which ad you’ll see. The ICCL argues that this isn’t just invasive, but also unlawful under the GDPR, as most users never give meaningful consent. Instead, they click ‘accept cookies’ on autopilot without realising their data is being sold in a global bidding war billions of times a day.
With Dublin serving as the EU base for many tech giants, our courts could soon become a key battleground for algorithm accountability. If the case succeeds, it could force platforms like Microsoft (and potentially TikTok, Google, and Meta) to rethink the recommendation systems that rely on personal data profiling.
Can We Blame It All on the Trend?
If the very architecture of these algorithms pushes users towards certain content, can companies really plead neutrality? Platforms argue that their algorithms are simply neutral tools, like a library index or TV guide—they don’t create content, they just sort it by connecting users to videos they are statistically likely to enjoy, just as a search catalogue connects readers to books. But unlike a library, TikTok’s engine is not passive – it is designed to maximise engagement, not neutrality. Its feedback loops can amplify extreme or sensational material because that’s what keeps users scrolling. From this view, responsibility lies with the creators of harmful material and the users who choose to act on it. If dangerous stunts or harmful misinformation continue to be promoted, that’s not neutral sorting – it’s a design choice with foreseeable risks.
So, is ‘the algorithm made me do it’ a valid defence? Right now, probably not for individual users. Courts are reluctant to excuse personal actions on the grounds of being influenced by a feed, but as lawsuits mount, one thing is clear: your FYP isn’t just entertainment, it’s evidence. So just remember the next time your FYP shows you a questionable trend, it could end up as Exhibit A.
Anna Portada O’Driscoll, BBL – Law Correspondent