Search

Type your text, and hit enter to search:

AI is rewriting the first hours of a crisis 

The first account of a crisis may no longer come from a journalist but from an AI-generated summary. Tony McChrystal explores the implications for reputation and why preparation must start long before a crisis hits.

Tony McChrystal 1 (2)
Image by Magnific

A recent CRJ article argued that trust is the basic infrastructure because it decides if a safety system works at all. The information that could prevent harm often exists long before it reaches the people who can act on it. 

Something similar is true in a personal or family crisis, with one new complication. What I increasingly see with families and executives is that when a principal is suddenly in the news, the first thing people read is no longer the coverage. It’s often a summary written by an AI tool, and the first hours of the crisis now begin there.

The clearest public example involved no family at all, but it shows the mechanism. Apple’s news notification feature condensed users’ alerts into AI-written summaries, and one of them, delivered under the BBC’s name, said that Luigi Mangione, the man accused of killing the UnitedHealthcare chief executive, had shot himself. That was untrue. Nobody had to open a report for the error to land, because it arrived as a notification carrying a trusted broadcaster’s logo. After a complaint from the BBC, Apple had, by January 2025, made its news and entertainment summaries temporarily unavailable.

I cannot say how quickly a given tool will absorb an early, inaccurate report, and I would be wary of anyone who claims to know. What the Apple episode shows is a change in the order of events. In a fast-moving story, a reader who meets only the summary may never reach the careful coverage that follows.

The harder problem is what happens afterwards. Publishing a correction is the familiar answer, and it matters, but it does not guarantee that a tool will change what it says. Consider two Munich publishers whose names Google’s AI Overviews linked to scams. According to the reporting, a formal demand that Google stop drew an inadequate response, so they went to court and won an interim order, which Google is free to challenge. The reported finding matters more than the order. The AI had drawn connections that none of the sources it linked to actually contained. An error like that cannot be fixed by correcting a source, because no source ever held it.

I draw two cautions from that. A letter, or even a court order, may not settle the matter, so a family cannot count on either. And where a tool draws on web sources, it stands to reason that accurate material is better than none, so the quality of what exists about a person matters before any crisis begins.

What should a crisis plan for an individual or family contain as a result? It should start with the first hours. Someone should be named in advance to ask the main AI tools about each family member in plain terms, as a journalist, banker, or counterparty might, and to keep a dated record of what each one says. That check should be repeated because answers change. 

If something is wrong, the accurate version needs to exist and be easy to find, which means a brief, plain statement of the facts, published quickly on a channel the family controls, worded so that it can be quoted without being misread, and updated as the facts develop. It helps if one named person owns that statement, so that the family speaks in a single voice.

The plan should also say who uses each provider’s routes for reporting errors, and how the response is recorded. Use whatever routes are on offer, but treat them as uncertain, and keep a record of each request and reply.

Finally, do the quiet work while nothing is wrong. Claim the person’s name, their domain, and their main professional and social profiles, and keep them accurate, so that when a tool goes looking it finds the person’s own account of who they are and not only whatever was said first. Nothing makes a tool draw on that material. But accurate sources give the family something to point to, and give the tools something better to find.

Tony McChrystal is Founder and Managing Director of Pavesen, advising executives, family offices, and high-net-worth individuals on online reputation.

    Tweet       Post       Post
Oops! Not a subscriber?

This content is available to subscribers only. Click here to subscribe now.

If you already have a subscription, then login here.