University of Bath researchers reported 455 UK-facing accounts offering to supply drugs through TikTok’s own in-app tool. Almost all reports came back cleared, every one of them after exactly 30.5 minutes.

Social media platforms publish large numbers about content moderation: millions of videos removed, the vast majority caught before anyone complained. A study from the University of Bath tested a different question. If an ordinary user spots an account that appears to be selling illegal drugs and reports it through the app, what happens?

The answer, published in the International Journal of Drug Policy, is that in 95.2 per cent of cases the report came back marked “no violation”.

“This is a straightforward test of a reporting system,” said Professor Chris Pudney of the Department of Life Sciences at Bath, who led the research. “These were accounts we identified because they displayed drug-related cues, targeted UK users and contained an explicit offer or route to supply. Yet 95% of our reports came back ‘no violation’.”

How the audit worked

Over five months, from February to July 2026, the researchers identified 455 TikTok accounts aimed at UK users that offered to supply drugs. To qualify, an account had to show a drug-related cue, be targeted at the UK, and contain an offer or a route to supply. The accounts referenced cocaine, synthetic cannabinoids known as spice, and controlled medicines.

The team then reported each one using TikTok’s standard in-app reporting tool, the same mechanism available to any user, and recorded what came back and what happened to the account afterwards.

Alongside the 95.2 per cent “no violation” rate, 83.5 per cent of the accounts were still publicly accessible 14 days later. Most of the changes in accessibility that did occur happened within the first two days; accounts that survived that window were unlikely to be removed later.

The pattern held across drug categories and across the different ways accounts promoted themselves. It was not confined to borderline cases of one particular type.

Exactly 30.5 minutes

One detail stands out. Every single “no violation” decision arrived precisely 30.5 minutes after the report was submitted.

“We could not determine whether decisions were made by automated systems, human moderators or a combination of both,” Pudney said. “However, the fact that every ‘no violation’ decision was returned at precisely 30.5 minutes suggests that at least some automation is likely to be involved in the process.”

Human review would be expected to produce variable response times, depending on queue length, the complexity of the case and the reviewer. A uniform interval, repeated hundreds of times, points to something running on a timer rather than a person looking at content.

What the law requires

The findings land in a specific regulatory context. Supplying or offering to supply a controlled drug is an offence under the Misuse of Drugs Act 1971. The Online Safety Act 2023 designates drug supply as a priority offence, and since 17 March 2025 regulated platforms have been required to take proportionate measures against priority illegal content. TikTok’s own Community Guidelines prohibit the trade and promotion of illegal drugs.

“The Online Safety Act treats offering illegal drugs for sale as a priority offence,” Pudney said. “These findings do not by themselves establish legal non-compliance, but they do raise an obvious question for Ofcom: are the systems and processes platforms have in place actually effective when users report this kind of content?”

Ofcom, the UK communications regulator, is responsible for enforcing the Online Safety Act. Its framework focuses on whether platforms have adequate systems and processes rather than on individual pieces of content, which is why the performance of a reporting pathway is a question within its remit.

Proactive numbers versus the report button

The researchers draw a distinction between two kinds of moderation. Proactive removal happens when a platform’s own systems detect and act on content. Reactive moderation happens when a user reports something. Platforms tend to publish figures for the former.

“It’s one thing for a platform to report very high proactive removal rates,” Pudney said. “It’s another for an ordinary user to point directly to apparent drug-selling accounts and be told, repeatedly, that there is no violation. Our findings show a serious weakness in the reactive reporting pathway we tested.”

“Platforms often publish impressive statistics showing how much content they proactively remove, but these tell us very little about whether a platform acts when a user directly reports apparent criminal activity,” he said. “That gap should be tested, not assumed away.”

The two measures can diverge in either direction. A platform might catch a great deal automatically while handling reports poorly, or the reverse. Published proactive figures alone do not reveal which.

The method the Bath team used has a name in the research literature: a prospective audit, in which researchers interact with a live system as ordinary users would and record what it does. The approach has been applied to housing discrimination, hiring and search ranking, and its value is that it measures behaviour rather than stated policy.

What the study does not claim

The researchers are careful about the limits of their work. The study was designed to test how the reporting system responded, not to estimate how common drug-selling content is on the platform, though they note that hundreds of eligible accounts were easy to find through ordinary searches.

They also say their findings do not prove legal non-compliance, and they acknowledge that TikTok reports large volumes of proactive removals. They could not establish who or what made the decisions they received.

The study is a single platform, a single reporting route, a single country and a five-month window. What it establishes is a measurement where previously there was mostly assertion.

A case for independent audits

The paper’s central recommendation is procedural: that regulators should test these systems themselves rather than relying on platforms’ own reporting.

“Ofcom should be independently testing whether the systems that ordinary users rely on to report suspected illegal content work in practice,” Pudney said. “If an ordinary user can repeatedly point a platform towards apparent drug sellers and receive ‘no violation’, that is exactly the kind of real-world performance a regulator should want to understand.”

The team calls for independent, offence-specific audits, the approach they used here, applied systematically. Testing a reporting pathway against one defined category of illegal content produces a measurable result, and the method could be repeated across platforms and offences.

Researchers from University College London and the University of Manchester also took part in the study.

---

Notes

1. Pudney, C., et al. “A prospective audit of TikTok’s response to UK drug-selling accounts.” International Journal of Drug Policy, 2026. DOI: 10.1016/j.drugpo.2026.105528.
2. University of Bath. “TikTok returns ‘no violation’ decision on 95% of reports of UK-facing accounts advertising drugs, study finds.” EurekAlert!, 20 October 2026. https://www.eurekalert.org/news-releases/1148490.

Photo by Craig Adderley on Pexels.com