Skip to main content

UK Online Safety Act for Adult Platforms: 2026 Compliance Guide

Regulatory · Deep-dive

UK Online Safety Act for Adult Platforms
2026 compliance, in plain English

The UK Online Safety Act 2023 became enforceable for "Part 5" services — services that publish or display pornographic content — in mid-2025. From January 2026 OFCOM is actively investigating non-compliant sites. Fines reach £18M or 10% of global revenue, whichever is greater. This guide is the operator-side summary: what you must do, what you can stop doing, and what the typical pricing looks like in 2026.

£18M
max fine
12 mo
audit-log retention
"highly effective"
age-assurance bar
Global reach
applies if any UK user
The trigger

What changed in 2026 — and why your VPN visitor doesn't help you

The Act applies to services that "have a significant number of users in the UK" or where the UK is a target market. There is no traffic-share floor: a site with 99% of traffic from the US can still fall under OFCOM jurisdiction if the operator markets to UK users, or if a single UK user can access pornographic content without highly effective age assurance.

OFCOM's enforcement playbook in the first six months: 1) traffic-sample audit on the largest 80 sites; 2) UK-source complaints route to enforcement queue; 3) formal information requests; 4) formal investigation; 5) penalty notice or business disruption order. The disruption order is the one operators dread — it instructs ISPs and payment processors to cut off the service.

The "VPN argument" — that all UK visitors are actually using VPNs — has been pre-rejected. Operators are required to take "proportionate" steps to detect VPN-based circumvention. Geo-blocking by IP alone is insufficient.

The requirement

"Highly effective age assurance" — what counts

OFCOM's August 2025 guidance lists six categories that qualify as "highly effective":

  • Photo ID + facial-similarity match — government ID upload plus live selfie matched within a defined similarity threshold
  • Facial age estimation — AI-based age estimate with a documented false-negative rate against under-18s
  • Mobile-network operator check — adult-content flag returned by UK MNO (EE, Vodafone, O2, Three) for the subscriber's account
  • Open banking-derived age — bank-confirmed age via Open Banking API
  • Credit-card check — credit-card-only verification (UK credit cards require 18+)
  • Digital identity wallet — UK Government Verify successor or accredited third-party (Yoti, Persona, AgeGo)

Self-declaration ("click here if you are over 18") is explicitly NOT highly effective. Cookie-based "remember my age" without re-verification at session start is also NOT compliant.

Audit obligations

What you must log and retain

WhatDetailRetention
Verification attemptsTimestamp, IP-hash, method used, outcome (pass / fail / fall-back)12 months
Method selection rationalePer-region / per-session reason the method was chosen12 months
Vendor identityWhich third-party processed the verification + their accreditation12 months
Override eventsAny case where a human operator overrode an automated fail5 years
Periodic-review recordsQuarterly OFCOM-style internal review of method efficacy5 years

Retention is from the last access to the record, not creation. Plan storage accordingly.

Real costs

What this actually costs a 100K-MAU adult platform in 2026

ItemOne-offAnnual
Yoti / Persona / AgeGo integration (Standard SDK)£5,000 – £15,000
Per-verification fees (assuming 30% return-rate, 100K MAU, 1 verify/quarter)£48,000 – £120,000
Audit-log infrastructure (storage + retrieval API)£3,000 – £8,000£1,200
Compliance officer time (0.2 FTE)£15,000 – £25,000
Periodic-review external audit£8,000 – £20,000
Total Year 1£8,000 – £23,000£72,200 – £166,200

Year 1 total range: £80K – £190K for a single-region UK-only implementation. Multi-region operators (UK + Texas + Louisiana + EU) typically 2-3× this. Cost driver is per-verification fee, not engineering.

FAQ

Common questions

Does the OSA apply to a site hosted outside the UK?
Yes. The Act is extra-territorial. If a UK user can access pornographic content without highly effective age assurance, the operator is in scope regardless of where the servers live.
Can self-hosted age estimation count as "highly effective"?
Only with documented false-negative rates against a representative under-18 population and periodic re-validation. Most operators use an accredited third-party because the documentation burden of "rolling your own" is higher than the per-verify fee.
What about the CCBill / Verotel angle?
CCBill updated their merchant requirements (v3.2, March 2026) to require OSA-compliant age assurance on UK traffic before processing. Verotel published similar policy in April 2026. Falling out of compliance puts your merchant account at risk before OFCOM gets involved.
Will Cloudflare / our CDN cut us off?
Cloudflare has historically been responsive to OFCOM business-disruption orders. The way to stay live is to fix compliance before the order issues — once it issues, getting the CDN back is a multi-week recovery.
Does a paid-subscription paywall count as age assurance?
Only if the payment instrument itself is 18+-gated (UK credit card √; PayPal ✗). Stripe + PayPal are mixed-age and do not qualify. SEPA direct debit is also insufficient.
What if a user fails verification — can they still browse a "PG" version?
Yes, and it is the recommended pattern. Send under-18 traffic to a non-pornographic mirror or a coming-soon page. OFCOM has explicitly approved the bifurcated approach as long as the gate decision is logged.