TL;DR

UK MP George Freeman confronted representatives from Meta, Google, and X before Parliament’s Science, Innovation and Technology Committee after an AI deepfake falsely depicted him defecting to the Reform party. None of the three platforms could offer a definitive commitment to removing such content, highlighting the gap between tech company policies and practical enforcement.

Three platforms, no clear answers

The exchanges before the committee revealed how differently each company approaches AI-generated disinformation — and how none of their frameworks produced a straightforward outcome.

Google’s Zoe Darme could not confirm whether the deepfake of Freeman would be considered “violative” under the company’s policies. X’s Wifredo Fernandez outlined a three-part test for evaluating such content, but acknowledged the framework applied generically rather than to the specific circumstances of Freeman’s constituency. Meta’s Rebecca Stimson said the fake had been “labeled by fact checkers” and “down-ranked,” reducing its engagement by 80 to 90 per cent — but the content remained on the platform.

For Freeman, labelling and down-ranking were not enough. He wants legislation establishing that “somebody’s identity belongs to them and cannot be stolen,” describing the deepfake as causing “serious disruption to democratic representation.”

Policy gaps under pressure

The hearing exposed a structural problem: platform content policies were designed primarily for user-generated content, not for AI-generated material that can fabricate political events with increasing realism. Fact-checker labels and engagement throttling assume that most users will see the warning, but research consistently shows that misinformation spreads faster than corrections.

Looking forward

Freeman’s call for identity protection legislation aligns with broader concerns across Westminster about AI-generated political content ahead of upcoming elections. The Online Safety Act addresses some categories of harmful content, but synthetic media that impersonates real politicians falls into a grey area that existing rules were not designed to cover. How quickly Parliament moves to close that gap may determine whether the next election cycle sees a significant increase in AI-generated political disinformation across UK constituencies.