Whether you agree with the Guardian’s conclusions or not, the underlying issue they’re pointing at is broader than any one company: the steady collapse of ambient trust in our information systems.
The Guardian ran an editorial today warning that AI companies are shedding safety staff while accelerating deployment and profit seeking. The concern was not just about specific models or edge cases, but about something more structural. As AI systems scale, the mechanisms that let people trust what they see, hear, and read are not keeping up.
Here’s a small but telling technology-adjacent example that fits that warning almost perfectly.
Ryan Hall, Y’all, a popular online weather forecaster, recently introduced a manual verification system for his own videos. At the start of each real video, he bites into a specific piece of fruit. Viewers are told that if a video of “him” does not include the fruit, it may not be authentic.
This exists because deepfakes, voice cloning, and unauthorized reuploads have become common enough that platform verification, follower counts, and visual familiarity no longer reliably signal authenticity.
From a technology perspective, this is fascinating.
A human content creator has implemented a low-tech authentication protocol because the platforms hosting his content cannot reliably establish provenance. In effect, the fruit is a nonce. A shared secret between creator and audience. A physical gesture standing in for a cryptographic signature that the platform does not provide.
This is not about weather forecasting credentials. It is about infrastructure failure.
When people can no longer trust that a video is real, even when it comes from a known figure, ambient trust collapses. Not through a single dramatic event, but through thousands of small adaptations like this. Trust migrates away from systems and toward improvised social signals.
That lines up uncomfortably well with the Guardian’s concern. AI systems are being deployed faster than trust and safety can scale. Safety teams shrink. Provenance tools remain optional or absent. Responsibility is pushed downward onto users and individual creators.
So instead of robust verification at the platform or model level, we get fruit.
It is clever. It works. And it should worry us.
Because when trust becomes personal, ad hoc, and unscalable, the system as a whole becomes brittle. This is not just about AI content. It is about how societies determine what is real in moments that matter.
TL;DR:
A popular weather creator now bites a specific fruit on camera to prove his videos are real. This is a workaround for deepfakes and reposts. It is also a clean example of ambient trust collapse. Platforms and AI systems no longer reliably signal authenticity, so creators invent their own verification hacks. The Guardian warned today that AI is being deployed faster than trust and safety can keep up. This is what that looks like in practice.
Question:
Do you think this ends with platform-level provenance becoming mandatory, or are we heading toward more improvised human verification like this becoming normal?
