On 28 August, the UK's product safety regulator published its newest annual report: cosmetics overtook electricals as the most flagged category, and 45% of the year's unsafe product notifications carry no risk rating at all

What the 28 August report actually covers

The Product Safety Database (PSD) is the register OPSS and local Trading Standards authorities use to log non-food consumer products found to present a risk, everything from a mislabelled cosmetic to a fire risk electrical appliance, drawn from the weekly reports market surveillance authorities issue when they find something. It excludes food and drink, vehicles, and medicines or medical devices, which run through separate systems. The 2025 to 2026 report is the fifth in the annual series, following editions covering 2021 to 2022 through 2024 to 2025, and it is the first to show cosmetics, personal care and grooming products overtaking electrical appliances and equipment as the most frequently notified category. In the prior reporting period, 1 April 2024 to 21 March 2025, electrical appliances led at 26%, followed by toys at 19% and cosmetics at 12%. In the newest report, that order has changed: cosmetics first, toys second, electricals third.

The numbers behind the report

All three figures come from OPSS's own published statistics for the 2025 to 2026 reporting period. None is an estimate.

The real subject: a register that cannot rank its own risks

Most coverage of the 28 August report led with the category swap: cosmetics overtaking electricals is a genuinely new finding, and an easy headline. The harder finding sits one line down. When a Trading Standards officer or OPSS itself logs a product on the PSD, entering a risk level, low, medium, high or serious, is optional. Nearly half of them skip it. That is not a rounding gap in an otherwise clean dataset; it means the UK's own central register of unsafe products cannot, for 45% of its entries, tell a business, a competitor, or a journalist how dangerous the flagged product actually was. The PORIYA electric scooter report published on 10 September, three days before this article, is a case where a rating was assigned, high fire risk, which is itself the more unusual outcome given the base rate the August figures describe. A market surveillance system that classifies risk on fewer than six in ten of the products it flags is not a system a compliance team can benchmark against with confidence: a category showing a modest notification count might be under-enforced, or it might simply be under-rated, and the published statistics alone cannot tell the two apart.

Why it matters for brands

Three groups carry the direct exposure. First, cosmetics, personal care and grooming brands: overtaking electricals as the most notified category for the first time is a signal that Trading Standards attention, and the testing and documentation demands that come with it, has shifted toward their sector, whatever the reason behind the shift. Second, any brand or platform bringing physical goods into the UK from outside it: import rejection at the border being the single most common corrective action, ahead of on-market destruction or marketplace delisting combined, means a large share of enforcement now happens before a product ever reaches a UK shelf or a consumer's door, at the point where customs and conformity paperwork is checked, not after a complaint. A brand that treats compliance documentation as something to have ready if a recall is ever requested, rather than something that has to travel with the shipment itself, is exposed at exactly the stage where OPSS's own figures say most enforcement now lands. Third, any team using the PSD, or the EU's comparable Safety Gate system, where cosmetics and toys have led the alert count for several years running, to gauge how risky a category is: the near half of PSD notifications with no risk rating means a low count in a category is not proof of safety, and a compliance function that reports category risk to its own leadership using OPSS's raw notification counts, without noting that classification gap, is repeating a data quality problem the regulator has not solved either.

Two ways to read the 28 August report

Frequently asked questions