Annual EU report on social media content moderation shows companies removed 62.5% of flagged material in March and April 2021, down from 71% in late 2019
social media giants like @facebook & @YouTube removed less hate speech in 2021 versus last year, according to leaked @EU_Commission audit coming later today. Read it here first 👉 https://y3r710.r.eu-west-1.awstrack.me/ ...
Context & Ripple Effects
This audit closes a five-year arc the EU itself has been publishing step by step: after the platforms' 2016 commitment to Brussels, they went from reviewing just 40% of flagged material within 24 hours (the Commission's first baseline) to removing 59% within a day (by mid-2017), 70% by early 2018, and 72% of illegal hate speech with 89% assessed inside 24 hours by February 2019 (the high-water mark).
The March–April 2021 figure of 62.5%, down from 71% in late 2019, is therefore not a one-off miss but the first reversal in the series — arriving just as Facebook's own transparency reporting claimed its enforcement had shifted heavily proactive. A leaked audit showing decline undercuts the voluntary Code of Conduct narrative the Commission has spent five years building.
First-order effects
- Facebook and YouTube face renewed public scoring against their own prior numbers, and the leak hands the Commission a concrete exhibit days before the official publication.
Second-order effects
- The reversal converts each successive EU report from a progress showcase into negotiating leverage, strengthening the Commission's position as it pushes the signatories from voluntary targets toward codified obligations with consequences for missing them.
- Platforms whose internal metrics show rising proactive takedowns must reconcile that story with the EU's user-flag-based measure, forcing changes in how transparency data is framed and reported.
Third-order effects
- If removal rates keep slipping while the audit cadence continues, the EU's experiment in self-regulated content moderation likely gives way to binding enforcement with penalties — turning an annual scorecard into a compliance regime.
The trend: Europe's voluntary hate-speech code is losing momentum after a five-year climb in removal rates, pushing the EU from annual scorecards toward enforceable platform-governance rules.