BBC Panorama traces overseas AI-fake networks pushing UK-decline narrative

TL;DR:

• BBC Panorama and the Top Comment podcast have identified dozens of interconnected Facebook and Instagram accounts running AI-generated videos depicting UK cities and society in decline, with operators based in Sri Lanka, the US, Vietnam, the Maldives, and locations linked to Iran and the UAE.
• The “Great British People” page (purporting to be Yorkshire-based but actually operated from Sri Lanka) drew 1.3 million views for an AI-generated video of an elderly white man crying about his pension.
• London Mayor Sir Sadiq Khan told the BBC that some operators are profit-motivated, others are backed by hostile states including Russia and China, plus US “MAGA”-aligned right-wing networks.

The most viral content includes AI-generated “2050” point-of-view walks through Liverpool, London, and Birmingham depicted as dirty, in flames, and dressed in Islamic clothing — content one account confirmed has surpassed 20 million views. The same operators also produced idealised AI-generated footage of Iran for comparison, with the creators describing themselves as based outside the cities they depict.

The detection gap

Research cited by Prof Yvonne McDermott Rees of Queen’s University Belfast finds the public detects AI fakes at roughly 55% accuracy — barely above chance — and consistently overestimates their own ability. Prof Sander van der Linden of Cambridge described the networks as “a new evolution of influence operations” supported by a growing “disinformation-for-hire industry” where paid actors and bots impersonate ordinary citizens to manufacture political support. Buying a pre-existing UK Facebook account to lend faux-British provenance is, he noted, cheap.

UK angle: a real-effect problem with no clean policy lever

Sir Sadiq Khan said City Hall research shows a sharp two-year rise in such posts and that the content has measurable effects — “putting off some visitors, overseas students or investors”. His policy ask is twofold: platforms amending recommendation algorithms so they do not reward divisive content, and clear labelling of AI-generated material. Meta said it takes “co-ordinated inauthentic behaviour seriously” but is enforcing against Community Standards rather than against AI-origin specifically.

Looking forward

The Online Safety Act’s “false information” provisions and Ofcom’s enforcement remit are the primary domestic tools, but neither addresses overseas-operated influence networks running through US-headquartered platforms. The next pressure points: whether OFCOM publishes guidance specific to algorithmic amplification of AI-generated political content, and whether the government’s AI Action Plan adds information-integrity workstreams beyond the existing AI Security Institute remit. For UK SMEs, the practical concern is that the same techniques can — and already do — target B2B reputation and procurement decisions.