Editorial Team
7 min read
Investigations and a TikTok search study found synthetic or impersonated clinicians reaching large audiences with dubious health claims. The evidence is not a platform-wide measure, but it shows why an AI label alone may not tell viewers whether a medical recommendation is credible.
TikTok, the short-form video platform that hosts and recommends these clips, is confronting a problem more specific than AI content in general: a health claim can arrive in the visual form of a doctor without the person on screen having made it. The evidence does not show how prevalent misinformation is across the service. It does show how search, impersonation and a professional-looking avatar can combine to make a claim appear more trustworthy than its source.
The clearest numbers come from a study by Hallam, a digital marketing agency. Its researchers found 235 videos with signs of AI generation or assistance among 1,198 publicly available TikTok search results across 25 high-stakes terms. In a separate search-based investigation, researchers identified 18 English-language accounts using AI-generated doctor videos within a few hours. A Spanish investigation found 70 profiles using manipulated images of real health professionals.
Company-reported Hallam analysis of selected TikTok health-search results showing signs of AI use. Source: Hallam.
Hallam created a new TikTok account to reduce personalisation, collected the top results for selected searches with Zeeschuimer, and manually reviewed videos for signs including synthetic voices, visual artefacts and repetitive scripts. Its sample was limited to publicly available top search results, not a random selection of all TikTok posts; it also grouped AI-assisted material with AI-generated videos and did not evaluate each video’s medical accuracy.
That method matters. The study’s 19.6% overall figure—235 of 1,198 videos—describes the selected result set. Health was its highest-AI category at 40%, while 42 of the first 50 results for “health tips” showed signs of AI use. The other four health searches produced sharply lower figures: 44% for “fitness,” 28% for “doctor advice,” 24% for “nutrition tips” and 18% for “mental health tips.”
| Measure in Hallam's selected sample | Result | What it does not establish |
|---|---|---|
| Videos with signs of AI use, all 25 searches | 235 of 1,198 (19.6%) | The share of all TikTok videos |
| Health-category average | 40% | The rate of misleading health information |
| “Health tips” top 50 | 42 videos (84%) | A typical health search or a platform-wide result |
| Average shares | 7,856 AI-signalled vs 7,221 non-AI | Why users shared, or a causal effect of AI |
The last comparison is a modest difference—about 9%—in average shares, while the AI-signalled group had lower average likes and comments. It is a distribution signal rather than proof that synthetic presentation caused sharing. Still, the examples reported alongside the research included disproven cancer myths, an invented anti-ageing product and home-remedy pitches. A report on the study says Hallam put the average views for the top AI doctor and health videos in its search at 2.5 million.
TikTok challenged the study’s representativeness. In a statement carried in that report, the company said findings from a marketing agency with commercial interests were not an accurate reflection of its platform, and said it removes harmful health misinformation, including AI-generated material, while working with the World Health Organization and the NHS on in-app reliable-health information.
The search study alone cannot show whether an avatar is impersonating a clinician or whether the health claim is false. The account investigations supply that missing, narrower evidence.
An investigation published in June reviewed 70 TikTok profiles selected through arbitrary keyword searches. The profiles used manipulated images of real doctors or pharmacists in videos that together had nearly 6 million followers. The investigators identified nine impersonated health professionals; 41% of the accounts did not display an AI-content notice. The videos appeared in English, Spanish, French, German and Japanese, with near-identical scripts sometimes paired with a different apparent clinician.
One example illustrates both the reach and the identity problem. The account dr.floriancaron had more than 199,000 followers, and one video had exceeded 27 million views as of June 17. Álvaro Fernández, a Spanish pharmacist and health communicator whose image appeared in the material, said he neither made nor authorized the false-health videos. The investigation also documented false claims about cancer symptoms, home remedies and skin treatments. Its arbitrary, search-led selection means it cannot put a total number on the practice.
An earlier English-language investigation found its 18 accounts by starting with “doctor advice” and following TikTok’s suggested-account feature. It identified weight-loss physician Garth Davis and orthopedic surgeons Paul Zalzal and Brad Weening—the hosts of the “Talking With Docs” podcast—among people depicted. Similar usernames, profile images and video designs may point to coordination, the investigators said, but copied scripts and reused videos mean the reports cannot establish how many operators were behind the accounts.
TikTok’s guidelines require creators to label AI-generated content and prohibit harmful impersonation. But a label establishes provenance, not whether an identifiable clinician gave the advice, whether a claim is evidence-based or whether a general recommendation applies to an individual.
That distinction was visible in the Spanish sample. Even when an AI label appeared, commenters asked the apparent doctor for help, thanked them or asked where to buy a product. Medical specialists consulted for the investigation said lab coats, technical language and an authoritative appearance can draw attention to the claim over a disclosure—particularly in rapid, mobile viewing. Those observations cannot measure how often labels fail, but they show why a technically present label is not the same as understood disclosure.
Clinicians and health bodies cited in the reporting see a potential public-health consequence. Prof Frankie Swords, NHS England’s national medical director, said false advice made solely for clicks was leaving patients worried and confused; Dr Emma Runswick, deputy chair of the British Medical Association council, warned against accounts that “peddle medical myths and promote so-called miracle cures.” Those are risk assessments, not evidence that every synthetic health video causes harm. The mechanism in the documented cases is more immediate: real medical authority is being copied onto claims the real professional did not endorse.
After the 18-account investigation was shared with TikTok, the platform said accounts violating its impersonation rule were removed. It also cited its third-quarter 2025 enforcement report: 99.3% of removed violative content and more than 98% of removed harmful-misinformation content were removed proactively. Those percentages refer to content TikTok removed, not to the share of all harmful content it caught, how many views preceded removal or the rate at which accounts returned.
The same distinction applies to money. Several accounts in the English-language investigation had follower counts far beyond TikTok Creator Rewards Program’s stated entry threshold of 10,000 followers and 100,000 video views in the previous 30 days; the examined videos were also over one minute long. That means they could qualify if they met the program’s other requirements. It does not show that they enrolled, were paid, shared an owner or were created for a single monetization plan.
The economic risk is therefore not a demonstrated revenue stream but a scalable route to audience-building. Synthetic videos and copied scripts can be reused across accounts and languages. If platforms act only after a specific post is reported, the relevant question is whether the same production pattern can rebuild reach before a removal takes effect.
The next useful evidence is not another headline percentage of AI video. TikTok would need to disclose results that connect its controls to exposure: time from upload to removal for health impersonation, views accrued before intervention, repeat-account rates and tests of whether labels change what viewers believe or do.
Until then, the evidence supports a practical limit on the reassuring story that AI labels solve the problem. A disclosure can identify a synthetic clip; it cannot authenticate a clinician, validate a medical claim or replace professional advice.
Enigma has emerged from stealth with a $71 million seed round and a public test involving at least 100 real robots. The startup says the interactions could improve robot interfaces and models, but it has not disclosed a specific commercial use case, customers or performance evidence.
Starbucks has retired Automated Counting, a computer-vision tool for counting milk and beverage components in North American coffeehouses. The company says it is standardizing inventory counts; reported miscounts leave the economics and operational value of the replacement unproven as Starbucks pursues more frequent replenishment.
HSBC plans to open a Global AI Centre of Excellence in Singapore in the second half of 2026 and hire more than 100 specialists. The bank is adding a deployment hub to an existing Google Cloud programme; the unresolved question is whether it can substantiate the promised gains while retaining human accountability in wealth, payments and treasury work.
Microsoft says MAI-Cyber-1-Flash lets its MDASH vulnerability system route routine work to a smaller in-house model and reserve GPT-5.4 for difficult cases. The resulting benchmark and cost claims apply to the combined system, leaving customer deployment, pricing and remediation outcomes to test in Project Perception’s preview.
A federal judge treated Anthropic’s purchased-book scanning and its model training as distinct fair uses, while leaving its earlier pirate-library copying exposed. Newly unsealed Project Panama records show how the company tried to replace that source of books at industrial scale.
Meta has raised its planned investment in the Hyperion campus in Richland Parish from $10 billion to more than $50 billion. Its promises of jobs, local investment and lower power bills now depend on tax terms and a 20-year utility arrangement whose long-run protections are still being tested.
AT&T has signed an agreement to explore D-Wave's annealing technology in more network operations after an early optimization workload dropped from about an hour to under 15 seconds. The commercial question is whether that company-reported result can deliver better decisions at the cost, scale and reliability of a live carrier network.
Yuyuantantian, an account linked to China Central Television, has proposed matching access to AI-model capabilities and risks rather than treating models as simply open or closed. The commentary is not a regulation, and separate reported discussions of limits on overseas access remain preliminary.
NVIDIA has made an undisclosed investment in Safe Superintelligence and promised access to its Vera Rubin platform that the companies say can increase the lab’s compute tenfold. The partnership gives Ilya Sutskever’s closed AI lab another hardware route, but leaves its research, capacity allocation and commercial terms unexamined in public.
NVIDIA has formed the Open Secure AI Alliance with more than 30 technology and security partners, using Hugging Face’s response to an AI-enabled intrusion as its case for deployable open-weight defensive tools. Members have identified work on identity, scanning and patching, but the coalition has not yet set out a common process for evaluating, disclosing or remediating failures.
CXMT's 466% Shanghai debut made the DRAM maker China's most valuable listed company, but the price was set with only 6.73% of its enlarged share capital freely tradable and before the company has proved how far its new capital can take it against tooling restrictions and established memory rivals.
Alphabet, Amazon and Meta have disclosed 2026 capital-spending plans that add to $520B to $550B at their stated ranges. The total conveys the scale of their infrastructure push, but it is not a comparable measure of AI-only spending or of the obligations each company is taking on.
Anthropic has made Claude Fable 5 generally available with classifiers that redirect certain risky requests to Claude Opus 4.8, while the same underlying model, Claude Mythos 5, remains available with cyber safeguards lifted only to approved organizations. The models share a listed token price, making vetting, monitoring and capacity—not a purchasable premium tier—the immediate constraint on less-restricted access.
Nvidia is reportedly discussing a roughly $250 billion financing backstop that could help OpenAI lease a proposed 10GW Ohio data-center campus. But no deal has been announced, other AI companies are pursuing the federally controlled power, and the reported 800MW first phase would be only a fraction of the planned buildout.
MSI's China distributor sheet shows week-over-week increases of roughly 8% to 20% for listed RTX 50 cards. Colorful's sheet shows substantial premiums to China MSRP, but without earlier prices it cannot establish how much—or how recently—those models rose.
Moonshot AI has released the weights and technical report for Kimi K3, its 2.8T-parameter mixture-of-experts model. The release gives researchers and operators a deployable checkpoint and more of its training infrastructure, but Moonshot still recommends large supernodes and its performance claims are not a uniform, like-for-like comparison.
xAI has added a built-in /deep-research command to Grok Build, turning a coding agent's existing delegation controls into a bounded, source-backed reporting process. The documentation is unusually specific about workflow limits and failure reporting, but it does not establish that the process improves accuracy, speed or cost.
An independently reported account says President Donald Trump posted AI-generated images of an Iranian tanker seizure, a burning tanker and a strike on Kharg Island. The posts did not announce a new operation, but they followed reported U.S. strikes on the island’s military assets and renewed attention to its oil terminal.
Alphabet’s proposed $80 billion equity package is partly tied to AI infrastructure, but its $40 billion at-the-market program is primarily intended to cover employee-equity tax obligations. Together with Microsoft’s $190 billion 2026 capital-expenditure forecast, the disclosures show a more complicated shift in how hyperscalers fund constrained compute capacity.
Intel is putting €5 billion into equipment and connections at its operating Leixlip campus to increase output of Intel 3-based Xeon processors. The investment strengthens an existing European manufacturing base, but Intel has not disclosed its added wafer capacity, the split between its own products and foundry work, or an outside customer.