Man used AI to fabricate complaints in bid to shut down London's Heaven nightclub

A London businessman has pleaded guilty to submitting fake complaints against the Heaven nightclub. According to the Metropolitan Police, he used artificial intelligence to generate the letters, which came from complainants who don't exist. It is the first publicly documented prosecution of this kind of scheme in the UK.

How the scheme worked

The defendant lived near Heaven, one of London's oldest and best-known LGBT+ venues, which opened in 1979. The club sits in Westminster and over the decades has become a cultural hub for Britain's LGBT community.

Under the UK's Licensing Act, any "interested resident" in the area can file a complaint about a venue, and the licensing authority must consider it. On paper, that's a sensible mechanism for democratic oversight. But that same requirement became the weak point: it assumes the complainant can be trusted, and it was never designed for a situation where the "residents" are generated by a neural network.

The man produced a series of complaints in the names of people who don't actually exist, complete with invented names, addresses in the right neighbourhood, and detailed accounts of "personal harm" from noise and the behaviour of the club's patrons. The sheer volume and apparent independence of the complaints were meant to simulate genuine public pressure and push the licensing authority into reviewing the club's licence.

The man pleaded guilty at a hearing; Heaven remains open.

Why it's hard to detect

A source in the Metropolitan Police said the Heaven case is not an isolated one. Police are seeing AI used to fabricate complaints from non-existent people more and more often, and it is already turning into a systemic problem for licensing authorities.

"Using AI to create letters from non-existent complainants is a growing problem," — a source in the

Metropolitan Police.

The main challenge is scaling verification. Checking a dozen complaints by hand is doable. But when a licensing authority receives a hundred letters from supposedly different residents, confirming they're genuine becomes practically impossible: staff would have to look up every name in public registers, track down the addresses, and try to contact each complainant. Most authorities don't have the resources for that.

The quality of modern text adds another layer. Unlike the template-driven mass letters of past years, which repeated the same phrasing, large language models produce stylistically varied content. Each letter reads like a real person wrote it, with its own quirks of speech, concrete details and personal stories. Spotting the forgery without specialised tools is extremely difficult.

Signs of manipulation

Investigators and licensing specialists point to several telltale patterns that help identify an AI-driven complaint campaign:

• a sharp spike in complaints right before hearings
• seemingly different authors whose arguments follow a similar structure
• names or addresses that don't appear in public registers
• complainants who don't show up at hearings and don't respond to inquiries
• all letters arriving within a narrow window after a review is announced

What it means

AI opens up a new avenue for abuse in administrative and licensing procedures: a single person can simulate mass public pressure with minimal time and money. Mechanisms originally built to protect residents become a weapon against competitors or unwanted neighbours.

UK lawmakers and licensing authorities will have to develop verification standards for digital complaints — otherwise the Heaven case will become routine rather than the exception. The scheme is technically viable in any jurisdiction where complaints are filed electronically and checked by hand.