The Warning Wore His Face: When a Company Deepfakes Its Own Founder

Warrier usually documents deepfakes deployed to deceive. This one was deployed to teach. At its 2026 annual meeting, Berkshire Hathaway opened the shareholder Q&A with a deepfake of Warren Buffett — made with no input from Buffett at all — and then revealed the trick to warn its own investors how easily anyone can now be faked. It is a rare, instructive case of an institution turning the weapon on itself, on purpose. And it raises a sharp question: when even the warnings wear a familiar face, what happens to trust?

What Happened

On 2 May 2026, at Berkshire Hathaway's annual shareholder meeting — the first with Greg Abel presiding after Buffett's transition to chairman — the Q&A session opened with a video of a figure introducing himself as "Warren from Omaha," asking, in Buffett's voice and manner, why shareholders should hold Berkshire stock for the long term. Abel answered, then told the room what it had just seen: the clip was a deepfake. Crucially, he added, it had been produced with zero input from Buffett — his appearance and signature voice replicated using only publicly available information — and he used the moment to warn shareholders about the growing threat that AI-fabricated content and cyberattacks pose.

The demonstration was pointed because Buffett is among the most impersonated people in fraudulent AI investment ads, the very "borrowed authority" scams that puppet trusted figures to sell fake schemes. Berkshire had even issued a rare press release on the problem, bluntly titled "It's Not Me." So the company chose to inoculate its audience by showing rather than telling — letting shareholders feel how convincing a fake can be, in a controlled room, before they meet the real thing in the wild.

Why It Matters

There is strong evidence that "prebunking" — showing people a manipulation technique in a safe setting so they recognise it later — works better than warnings alone. Berkshire's stunt is that principle applied at scale to the people most targeted by Buffett impersonations: its own investors. That makes it a genuinely constructive use of the technology, and potentially a model. But it also sits on a knife-edge. A trusted institution fabricating its living founder, without his input, to teach a lesson is powerful precisely because it is unsettling — and everything that makes it effective also makes it risky if the guardrails slip. The line between a warning deepfake and a weaponised one is a single missing sentence: "that was fake."

advertising