When Every Video Might Be Fake, the People Who Know It Best Are Begging You to Stop Sharing

A post on X this week cut through the usual arguments about AI and misinformation with something rarer than data: moral urgency. The account @Ahmed04Younis, replying to someone who had shared what appeared to be footage of suffering in Gaza, wrote: "Please don't spread misinformation, this video is AI. We have suffered enough and we have more real horrendous footage. We will lose our honesty because of this fake stuff." The post got 111 likes and 7 retweets — modest numbers in the attention economy, but the kind of post that carries weight disproportionate to its reach, because it names the actual stakes. Not "deepfakes are a problem." Not "AI threatens trust." Something more specific: the people with the most to lose from fabricated atrocity footage are the witnesses trying to document real ones.

The same 48-hour window produced a different kind of complaint — less anguished, more damning. A Bluesky post with 433 likes declared that Google has become "perhaps the largest source of misinformation in the world," adding that a specific result it had served up was "just completely made up." The two posts are doing different work. The Gaza post describes a crisis of credibility in a conflict zone where documentation is survival. The Bluesky post describes something more diffuse — a slow erosion of the basic expectation that a search engine returns facts. Together, they bookend the AI misinformation problem in a way that institutional coverage rarely manages: there's the catastrophic version, where fabricated footage delegitimizes real testimony, and the ambient version, where minor hallucinations quietly degrade the information environment for everyone.

What makes this moment distinct from earlier cycles of misinformation anxiety is the direction of the critique. A third post, circulating on X, summarized research from Perle Labs arguing that faster AI means faster misinformation — specifically that content creation speed has outrun verification infrastructure. This is the structural diagnosis. But the structural diagnosis, accurate as it is, doesn't capture what the Gaza post captures: that the communities documenting the most consequential events are the ones now spending energy fighting fakes that undermine their real footage. The burden of proof has inverted. You used to need evidence to be believed. Now you need evidence that your evidence isn't fabricated.

The mood across the conversation shifted noticeably in the last day — less catastrophizing, slightly more analysis — but that softening shouldn't be mistaken for resolution. It reflects the news cycle moving on, not the problem shrinking. What the University of Edinburgh's deepfake research confirmed this week — that AI fingerprints can be stripped from generated images, making origin detection essentially impossible for most viewers — means the verification gap Perle Labs described isn't a temporary bottleneck. It's the permanent condition. The witness in a conflict zone who says "don't share this fake, here is the real footage" is now doing a job that no detection tool, no platform policy, and no AI watermarking standard can do for them. The people with the least institutional power have become the last line of defense against the technology that the most powerful institutions built.