Ofcom on Friday opened a consultation proposing that category 1 platforms — Facebook, Instagram, Snapchat, X, YouTube, Google and ChatGPT among them — be legally required to ban scam advertisers from their services, block bad actors from creating new accounts after being caught, and give law enforcement dedicated channels to flag fraudulent ads The Guardian.

The draft measures, issued under the Online Safety Act, would also require platforms to reduce the risk of accounts being hijacked for scam purposes and to ensure financial-services advertisements carry proper legal clearance before they run. Oliver Griffiths, Ofcom's online safety group director, said tech companies "had not done enough" to combat fraudsters operating on their platforms The Guardian.

The consultation runs until October 2026. Once the specific steps are finalised and become legally binding, non-compliant platforms face fines of up to 10% of global revenue — the Online Safety Act's headline enforcement mechanism. But Ofcom will not issue final decisions until 2027, meaning any binding requirement is at least a year away from taking effect.

That timeline has already drawn criticism. Which? welcomed the proposals but flagged the delay as a real problem. Rocio Concha, the consumer group's head of policy and advocacy, argued that advances in AI are making scams more sophisticated in the interim, and that a year-plus wait leaves consumers exposed to fast-evolving fraud techniques while regulation catches up The Guardian.

The scale of the underlying problem is not in dispute. The Online Safety Act's own impact assessment puts the annual economic and social cost of online fraud in the UK at £18.9 billion Ofcom consultation response. Since 17 March 2025, providers covered by the Act's illegal content codes have been required to take the safety measures set out in those codes, or adopt equally effective alternatives Ofcom — but that duty has not stopped scam ads proliferating on major platforms.

Meta's record illustrates the gap between pledge and practice. The company promised to stop illegal financial ads appearing on its platforms in Britain, but a Reuters investigation in March 2026 found such ads still running roughly 1,000 times a week Reuters. Ofcom acknowledged at the time it remained powerless over paid-for scam ads specifically because the relevant statutory provision had not yet come into force — a gap Friday's consultation is designed to close.

Meta's relationship with the regulator has grown more adversarial rather than more cooperative. In May 2026 the company filed a legal challenge against Ofcom over online safety fees and fines Reuters, a dispute that sits alongside — but separate from — the scam-ad consultation. Ofcom has also singled out TikTok and YouTube for lagging on child-safety measures relative to competitors, reporting in May 2026 that both platforms had failed to set out meaningful steps to protect British children from harmful content while rivals had acted Reuters. Taken together, the pattern suggests a regulator repeatedly finding categories of platform behaviour it judges inadequate, well before its own enforcement powers are fully switched on.

The AI dimension gives Concha's warning particular weight. The Bank of England cautioned the public in June 2026 against AI-generated scams after deepfake videos depicting Nigel Farage in a physical altercation with its governor circulated online The Guardian. Financial commentator Martin Lewis has for years pressed the government to act against scam advertisements that misuse his image and likeness to lend fraudulent schemes credibility — a long-running grievance that predates this consultation but that the new proposals are explicitly aimed at addressing.

The scam-ad measures arrive alongside a separate set of draft rules covering journalistic content and harmful material more broadly. Ofcom is proposing protections against arbitrary removal or downranking of news content, requiring platforms to give publishers a chance to respond before action is taken. It also wants category 1 platforms to limit user exposure to content involving suicide, self-harm, eating disorders and hate or abuse, and to give users tools — including the ability to block or mute accounts and filter out interactions with unverified accounts — to manage their own exposure The Guardian.

The bundling of scam-ad, journalism and harmful-content rules into a single consultation round reflects how the Online Safety Act's category 1 regime is being built incrementally, provision by provision, rather than switched on wholesale. Each strand carries its own enforcement timeline and its own industry pushback, which is part of why critics see the phased approach as structurally slow relative to the pace at which fraud techniques, particularly AI-assisted ones, are evolving. Whether the eventual fining power proves a genuine deterrent will depend largely on how Ofcom uses it once 2027 decisions are made — and on whether platforms with the resources to litigate, as Meta has already shown a willingness to do, treat compliance as cheaper than contesting it.