TikTok EU and UK trust-and-safety staffing substitution

Every video, photo and text item uploaded to TikTok passes a real-time automated review before anyone but its creator can see it. TikTok's published Year 3 systemic risk assessment describes the machinery in its own words: detection runs on "vision-based, audio-based, text-based and LLM-based" technologies plus keyword lists and natural-language processing, automated removal is "applied when violations are the most clear-cut", and otherwise the item is routed to a human queue. High-view content can be pulled back for additional human review. Specialist queues exist for misinformation, worked against a repository of previously fact-checked claims from accredited fact-checking partners, and for illegal content, advertising and marketplace listings. Every decision is appealable to a human and then to an external out-of-court dispute settlement body. No AI model is named anywhere in the record and none is named here.

The scale is published because the law compels it. In the six months to June 2025 that apparatus removed 24,534,707 pieces of content under the Community Guidelines, restricted 169,527,678 more, imposed 2,781,470 service restrictions and 4,906,735 account bans or suspensions, of which 871,819 were automatic. It handled 3,075,758 appeals from uploaders and advertisers and 1,054,432 from users who had reported content — about 22,700 appeals a day — and reinstated or unrestricted 1,359,823 pieces of content, at a median decision time under two hours on both tracks. It received 308,755 illegal-content reports from EU users covering 151,354 unique items, of which 26,512 were actioned as unlawful and 15,365 as policy breaches; 82 trusted-flagger reports; and 3,976 orders from Member State authorities, actioned at a median under three hours. Behind all of it sat 169 million monthly active recipients in the European Union, up from the 135.9 million declared when TikTok was designated a very large online platform in April 2023.

The number of people doing the human half moved the other way. TikTok's second DSA transparency report recorded "more than 6k" moderators dedicated to EU content at the end of December 2023. Its fifth report gives 4,596 at the end of June 2025, including 247 who are not language-assigned, with a per-language table running from 1,552 English down to 63 Swedish, 11 Latvian, 10 Croatian, 10 Estonian, 5 Lithuanian, and 0 Irish and 0 Maltese. Its sixth report, as read by EUobserver, gives 91 employed staff against 3,583 contracted human moderators. Whole sites went in between: the entire 300-person Netherlands moderation team in September 2024, fewer than 500 Malaysian roles in October 2024, about 300 Dublin trust-and-safety roles notified in March 2025, about 150 Berlin trust-and-safety and TikTok Live roles announced on 10 August 2025, approximately 430 London roles put at risk with notices on 22 August 2025, and about 300 further Dublin roles proposed on 1 July 2026. Social Media Today's arithmetic over the same disclosures puts the audience up about 25 percent since September 2023 and the moderation workforce down about 26 percent; that is an analyst's computation over TikTok's own figures, not a TikTok statement and not an audited one.

What makes the London reduction unusual is TikTok's own account of what it removes. Writing to the Chair of the House of Commons Science, Innovation and Technology Committee on 7 November 2025, its Director of Public Policy stated that "In the UK, there are approximately 430 roles at potential risk under this proposal", and then broke them down: "a third of these roles are in teams involved in the labelling of data for AI model training. Progress in the development of these models has significantly decreased the need for this kind of manual labelling. Another significant proportion of those potentially affected are in ancillary roles, for example training teams, whose duties include activities such as training moderators on our Community Guidelines... Around a third of those impacted are front line moderation teams." So two of the three functions in the proposal are not the judgement itself but the two loops that maintained it: the people who produced the training signal the classifiers learn from, and the people who calibrate the reviewers who produce that signal.

The same letter describes a re-partition as well as a reduction. TikTok said it is "moving from a region-based structure of generalised moderators to one based on different types of products or types of risk, known as verticals" — harassment, misinformation, fraud — consolidated into fewer sites and supplemented by third-party specialists offering "greater ability to rapidly expand" and "greater levels of language-specific, follow-the-sun coverage", and that the arrangement is "not a like-for-like replacement". The language-indexed staffing table the DSA requires TikTok to publish therefore describes a partition the operator says it is leaving.

Now the regulator, which is where this case stops being an employment story. On 19 February 2024 the Commission opened formal DSA proceedings against TikTok covering protection of minors, advertising transparency, researcher data access, and the risk management of addictive design and harmful content, naming Articles 34(1), 34(2), 35(1), 28(1), 39(1) and 40(12). On 17 December 2024 it opened a second, separate proceeding on election-integrity systemic risk after the annulled Romanian presidential first round, limited to recommender systems and political advertising, with Coimisiún na Meán associated to the case. Neither proceeding names content-moderation staffing. As of 28 August 2026 the first has produced four sets of preliminary findings — the advertisement repository on 15 May 2025, researcher data access on 24 October 2025, addictive design on 6 February 2026, and minors' account settings on 24 July 2026 — and one closure by binding commitments, on advertising transparency, on 5 December 2025. An earlier proceeding on the TikTok Lite Rewards programme, opened 22 April 2024, closed on 5 August 2024 when the Commission made binding TikTok's commitment to withdraw the programme from the EU permanently: the first DSA case closed and the first commitments accepted. There is no non-compliance decision and no fine. Preliminary findings are not findings of breach, and TikTok said of the addictive-design set that "The Commission's preliminary findings present a categorically false and entirely meritless depiction of our platform, and we will take whatever steps are necessary to challenge these findings."

The staffing question was put to the Commission directly. On 6 November 2024 Kim Van Sparrentak asked, in written question E-002454/2024, whether a platform can comply with DSA Articles 16, 20 and 34 to 35 after firing an entire national moderation team of 300 people. The Commission's answer, last updated 15 January 2025, is the load-bearing sentence of this whole case: "The DSA does not prescribe any specific rules about the resources to be dedicated to content moderation." It went on to say that platforms must enforce their rules "in a diligent, objective and proportionate manner", that "qualified staff" must ensure fair and unbiased decision-making in internal complaint handling, that "it is important that designated companies put in place adequate content moderation processes and dedicate enough resources for diligent content moderation", and that it "is closely monitoring TikTok's compliance with the DSA and will follow up with formal enforcement steps if appropriate". No such step naming staffing has followed.

So the pressure came from elsewhere. In Berlin, ver.di held five one-day strikes in July 2025 and a four-day strike from 23 September 2025, demanding a collective agreement with severance worth three years' salary and a twelve-month notice extension, under the slogan "We trained your machines, pay us what we deserve!"; TikTok said the changes would "streamline workflows and improve efficiency" while maintaining its commitment to safety, and dismissal cases went to the Berlin Labour Court with no reported outcome. In London the redundancy notices of 22 August 2025 landed days before a scheduled ballot on union recognition with UTAW, a branch of the Communication Workers Union; on 19 December 2025 two moderators, supported by Foxglove and UTAW and represented by Leigh Day, sent a pre-action letter alleging unlawful detriment and automatic unfair dismissal. TikTok told Parliament the union-timing claims are "categorically untrue", that the decisions were made globally, and that it had written to the CWU expressing "regret about these timescales" and remained open to re-engaging after consultation. These are allegations and denials. No tribunal has ruled.

The parliamentary limb produced the most. After a Trades Union Congress open letter, TikTok wrote to the Committee on 20 October 2025; the Chair, Dame Chi Onwurah, replied on 28 October with six questions, including whether a risk assessment of the job losses for UK user safety had been conducted and what its outcome was, and quoted TikTok's own earlier evidence back to it — "tens of thousands" of safety professionals working 24/7, a 2024 trust-and-safety investment of over $2 billion, and oral evidence of 25 February 2025 distinguishing what automation handles well, "pornographic material, blood and that kind of thing", from content referred to "human moderators who have to use their nuance, skills and training to be able to rule on other elements that can include hateful behaviour and misinformation". TikTok's 7 November reply described an internal analysis expecting improvements in the speed and the efficacy of moderation, and supplied no data. On 13 November the Committee published the reply under the headline that TikTok had failed to share the evidence. The Chair's comment: "TikTok have come up empty to show that this transition to AI won't lead to more harms for its users."

That leaves the measurement, which is mandated, self-graded, restated and independently qualified all at once. TikTok's automation figures rise across the period on scopes that widen underneath them: 72.3 percent of Community Guidelines removals were automatic in the January-June 2025 DSA report; "86% of the content we remove is now removed by automation", TikTok told Parliament in October 2025, citing an April-June 2025 report from a different family; and in the July-December 2025 report "Automated systems actioned 93.8% of all violating content without human review", in the first period to include comment enforcement. TikTok itself cautions that each edition captures a broader range of automated enforcement actions than the last, so the three figures are three separate statements and never a trend. Its accuracy figure is an overturn-rate complement: accuracy is defined as the proportion of decisions "upheld or maintained" and error as the proportion "overturned", reported at 99.2 percent and 0.8 percent for automated moderation in the first half of 2025, with per-Member-State error running from 0.3 percent in Slovakia and Bulgaria to 1.6 percent in Austria. In the same six months 3,075,758 uploader appeals produced 1,359,823 reinstatements — a signal an order of magnitude larger on a different denominator, which TikTok warns does not map one to one onto the period's appeals.

The independent audit the DSA itself mandates then took a view. The audit firm KPMG Advisory N.V.'s Article 37 assurance report for the year to 30 June 2025, dated 29 August 2025, returns a NEGATIVE, qualified opinion across 90 specified requirements. Four of its conclusions are negative, and all four sit inside the moderation record: on Article 16(6) it identified notices where no moderation action was performed and could not evidence the monitoring controls over the interface between notice intake and the moderation systems, so it "could not obtain sufficient assurance to support the completeness of the total population of notices"; on Article 20(4) complaint records could not be retrieved "due to limitations in documentation retention", so it "was unable to confirm that all complaints were handled in a timely, non-discriminatory, diligent, and non-arbitrary manner"; on Article 24(5) duplicate statements of reasons were transmitted to the Commission's DSA Transparency Database, producing more records than the actual number of decisions taken, until a May 2025 remediation, and post-remediation sampling still found statements of reasons that were never transmitted; and the advertisement repository was found defective under Article 39(3). Six requirements were disclaimed because they sit under the Commission's open proceedings. On Article 42(2) — the article that mandates the per-language headcount and the accuracy indicators themselves — the conclusion is "positive with comments", the comment being that "internal controls concerning data accuracy and completeness monitoring, between the various source systems and Transparency Report are not sufficient and appropriate". On 15 April 2026, nearly eight months after publication, TikTok restated Annex D of its January-June 2025 report, noting that the values "have been updated with the correct values". The mandated record of how many people moderate in each EU language was wrong when it was first published.

One document is conspicuous by what it does not contain. TikTok's published Year 3 systemic risk assessment is dated 28 August 2025 — eighteen days after the Berlin announcement and six days after the London notices. It records the Fundamental Rights inherent risk as Medium-High and Likely, "consistent with TikTok's score in Year 2", and names over-moderation and under-moderation by "content moderation systems and human moderators" as a standing moderation risk. It contains no reference to the trust-and-safety workforce reduction, restructuring or redundancies. That is an observation about a published text, not proof that the matter was never assessed internally: the document is headed confidential and this is its public version.

And no independent measurement of the outcome exists. There is no published study of what the staffing substitution did to moderation quality, in any language, on any policy. The available signals are TikTok's own upheld-and-overturned statistics, an auditor who could not confirm the completeness of the populations those statistics come from, and a Commission preliminary finding of 24 October 2025 that TikTok and Meta may have put in place burdensome procedures and tools for researchers to request access to public data, "often leav[ing] them with partial or unreliable data". The parties best placed to check the claim from outside are the ones the regulator says cannot get reliable data.