Big Tech’s moderation mistakes are piling up in Europe’s appeals system

(CN) — In Europe, banned posts on platforms like TikTok, YouTube and Instagram are getting something Silicon Valley hates: appeals.

A new report from Dublin-based Appeals Centre Europe suggests those appeals are working. The body said major platforms, including Meta’s Facebook and Instagram, TikTok, and YouTube, repeatedly left up hateful or policy-violating content while also wrongly deleting legitimate posts and suspending accounts across the European Union.

Created under the EU’s Digital Services Act, the center lets users challenge moderation decisions outside the platforms’ own internal systems. Between April 2025 and March 2026, it received more than 24,000 complaints, with over 12,000 qualifying for review. Eligible disputes exploded over the year, rising to “nine times as many eligible cases in March 2026 than we did in April 2025.”

Hate speech produced some of the starkest numbers. After reviewing more than 1,400 cases where platforms refused to take down reported content, the center overturned those decisions 70% of the time. TikTok posted the highest reversal rate at 83%, followed by Instagram at 74%, Facebook at 61% and YouTube at 58%.

Reviewed posts ranged from antisemitic and anti-migrant content to racist abuse, anti-Roma speech and attacks targeting LGBTQI+ communities.

“Online hate and harassment have real-world consequences for many people and communities,” Thomas Hughes, chief executive of the center, said in comments accompanying the report. “In more than two-thirds of our decisions about hate speech, we found that platforms failed to enforce their own policies and left up hateful content.”

Moderators also struggled with scams, manipulated media and misinformation. One AI-generated TikTok video about the Russia-Ukraine war used a fake voiceover layered onto a real news broadcast. The center concluded the clip violated TikTok’s misinformation rules, but said it was still online weeks later.

Racist Instagram comments posted after a Liverpool-Galatasaray Champions League match compared Black players to monkeys. The center said all 20 reviewed comments violated Meta’s hateful conduct rules, yet several remained online afterward.

Moderation mistakes were not limited to harmful content left online. The center overturned platform decisions in 52% of disputes involving removed posts after reviewing the underlying material. One case involved a Czech photographer whose nude artistic image was removed from Facebook even though the nipples had been blurred to comply with Meta’s policy. The center recommended restoring the image with an age restriction, and Meta later reinstated the post.

The largest flood of complaints came from suspended accounts and restrictions. More than 14,000 users sought reviews of bans or limitations, with over 5,000 disputes considered eligible. But platforms rarely handed over the material needed for independent review. By the end of March, the center said it had received enough evidence to fully assess fewer than 150 account-suspension cases.

Instead, it issued more than 7,000 “default decisions,” procedural rulings automatically handed to users because platforms failed to provide the requested evidence.

“This is currently leading to a large number of default decisions, where a platform does not provide the content for an eligible case and we make a procedural decision in the user’s favour,” the report said.

Hughes said the group is beginning to spot recurring moderation failures across thousands of disputes, though he stopped short of calling the platforms systematically noncompliant with the Digital Services Act.

“Appeals Centre Europe is happy to present data that points to patterns and areas for improvement, but it is for the relevant regulators to determine when a matter becomes an issue for systemic compliance,” Hughes told Courthouse News.

He said platforms routinely fail to cooperate when users appeal account suspensions, often blaming “operational limitations” instead of sharing the evidence needed for independent review. The center urged companies to overhaul those systems so users can actually exercise their new EU appeal rights.

Hughes said many moderation failures appear tied to weak contextual review, limited language expertise and users deliberately trying to game automated detection systems.

“Platforms make regular and frequent mistakes in their content moderation and the practical effect is that harmful and policy-violating content remains visible to users,” he said, pointing to tactics such as replacing slurs with symbols or mixing alphabets to evade moderation filters.

Participation varied sharply across Europe. France generated the highest number of eligible disputes, followed by Belgium, Italy, Spain and Germany. On a per capita basis, Belgium ranked first, followed by Lithuania, Cyprus, Malta and Slovakia.

Civil society organizations accounted for a large share of the complaints, particularly in hate speech disputes. Slovak group digiQ, Polish anti-racism organization NEVER AGAIN Association and Belgium-based Media Diversity Institute Global submitted large batches of complaints involving racist, antisemitic and xenophobic content left online by platforms.

The center’s decisions are nonbinding, meaning platforms are not legally required to restore content or remove posts simply because reviewers say they should. But the Digital Services Act does require companies to engage with certified dispute-settlement bodies “in good faith,” and repeated refusals to provide evidence or participate meaningfully could increasingly attract scrutiny from national regulators and the European Commission, which oversees enforcement for the bloc’s largest platforms.

For years, getting banned, suspended or ignored on social media often meant arguing with an algorithm and hoping someone at the company noticed, with users around the world largely limited to opaque in-house appeals systems run by platforms like Meta, TikTok and YouTube. Europe is now trying something Silicon Valley long avoided: giving ordinary users a real way to fight moderation decisions and sometimes win.

Courthouse News reporter Eunseo Hong is based in the Netherlands.