LO
DeploymentCognitive automation2026-06-30
TikTok's statutory DSA report for H1 2026 says human moderators feed corrections back into its moderation models, while technology handles clear-cut cases and people take new or complex ones
Content moderatoroccupation page →
Event date / reported
2026-06-30 · reported 2026-08-31
Evidence stage
DeploymentAn employer has put it into production. Can move the baseline — weighted by scale and how similar the setting is.
Tasks this bears on
The call the rule does not decide
Satire, reclaimed slurs, a war video that is evidence, a medical image that is not what it looks like — the item where applying the rule literally gives the wrong answer.
Still human-led✓ Evidence-backed
Correcting the system that replaced the queue
Labelling the cases it got wrong, finding the pattern behind a cluster of bad calls, and being the person who can say why a rule and a model disagree.
New task✓ Evidence-backed
Where this applies
The platform's own statutory report under the EU Digital Services Act, in the Commission's template, for 1 January to 30 June 2026. In its qualitative section it says its moderators work alongside automated systems and take into account context and nuance that technology may miss; that human moderation also helps improve the automated systems by providing feedback for the underlying machine-learning models; and that its approach allows technology to handle clear-cut cases while human moderators focus on new or complex issues. The same template counts 3,738 moderators with EU-language expertise, 3,676 of them external. It is the company describing its own system under a legal duty to report, with a stake in appearing careful; it gives no measure of how much moderator time goes to feedback for the models.
What this means
On a platform this size the machine takes the obvious cases and the people take the new and the ambiguous — and the people's corrections are what the machine learns from. Teaching the system is part of the moderator's job, described by the platform itself.
What it does not yet show
It is the platform's own description under a reporting duty; it does not measure how much moderator time goes to feedback, or whether the complex share is growing.
What you can check
Open TikTok's DSA transparency report for January to June 2026 (statutory template) and find, in the qualitative tab, "Human moderation also helps improve our automated moderation systems".
Does it change the assessment?
No. The impact index is never moved by a single event. What this record did: the 2 linked task judgements above now rest on evidence instead of inference.
Source
TikTok — DSA Transparency Report, January to June 2026 (statutory template; date of publication 2026-08-31) · verified 2026-09-27 · Claude (VOLO agent) · interpreted 2026-09-27 · Claude (VOLO agent)
Primary source — published by the party that did this, or the authority of record. No co-signature needed.