Can You Trust an AI's Answer About Who Runs a Company?
Mathijs Bronsdijk5 min readUpdated October 8, 2026
On this page
On August 31, 2026, UPS announced that six executive vice presidents would report directly to CEO Carol Tomé under a new global operating model, effective the next morning. One 37-year veteran retired from her post.
A new Chief Global Commercial Strategy Officer role was created and left open, “a search is underway.” The filing is public, dated, and free to read.
No AI model available today was trained on it. Ask one who runs UPS’s international business and, unless it happens to search the live web first, it will answer from whatever it last learned, with nothing in the response to flag that the answer might already be six words too old to be true.
What UPS actually announced
The filing names names. Kate Gutmann, EVP and President of International, Healthcare and Supply Chain Solutions, is retiring for personal reasons; her final day in the role was September 1.
She’s succeeded by Wilfredo Ramos, a 20-year UPS veteran who previously led the company’s Asia Pacific and brokerage businesses, now titled EVP and Chief International, Healthcare and Supply Chain Solutions Officer.
Two more roles were created outright. Nando Cesarone becomes EVP and Chief Global Operations Officer, overseeing the company’s air network, ground operations, and engineering. Matt Guffey becomes EVP and Chief U.S. Domestic Officer, running Small Package, Roadie, Happy Returns, and the UPS Store network.
A seventh seat, Chief Global Commercial Strategy Officer, is named but unfilled: UPS says a search is underway for it right now.
That last line is the kind of signal our own opportunity-signals guide reads for: a public company naming an open seat in its own press release, before a firm ever gets a call about it.
Watch a client's market continuously, not once a quarter.
Book a callWhy this isn’t the same problem as a hallucination
Our checklist for catching AI-hallucinated research covers a different failure: a model inventing a citation, a quote, or a source that never existed. That’s a fabrication problem, and the fix is tracing every claim back to a real document.
Staleness is quieter than that, and in some ways harder to catch. The model isn’t lying. “Kate Gutmann leads UPS International” was a true, verifiable fact for six years. Nothing about the sentence looks wrong.
The model has no internal signal that says “this was true as of my training cutoff, and roles like this change.” A language model doesn’t know what today’s date is unless something tells it, and it doesn’t track which of its facts have expiration dates.
Cluen’s most recent survey of search firms’ technology use, covered in our piece on how search firms use AI, found 73.5% of firms use ChatGPT, most often to draft candidate write-ups and job descriptions. None of that adoption changes how training cutoffs work. A firm using AI to draft a target list is drafting from a snapshot, and every snapshot ages the moment it’s taken.
Where a stale seat actually breaks a market map
Three places, in order of how often they get missed.
A candidate’s current title
A passive candidate you’re tracking gets promoted, moves companies, or changes roles, and an AI-assisted profile that hasn’t been re-checked against a current source still shows the old one. The pitch to a client references a seat the candidate no longer holds.
A client’s own leadership roster
If UPS were your client, “who does Ramos’s old Asia Pacific role now” is a question your own briefing document needs to answer correctly on September 2, not whenever a model was last trained.
A comp benchmark’s peer set
Executive compensation benchmarking leans on named public-company comparables. If one of those comparables just restructured the exact role you’re benchmarking against, the peer figure is now describing a job that doesn’t exist in that form anymore.
How to verify who actually holds a role today
| Source | What it confirms | How current |
|---|---|---|
| Company newsroom / investor relations page | The company’s own announcement | Usually same-day |
| SEC 8-K, Item 5.02 | The filed, legally required record of the change | Within 4 business days of the event |
| Recent DEF 14A proxy statement | A full snapshot of the named executive team | As of the filing date, can lag a fast-moving reshuffle by months |
| A general-purpose AI model, unprompted | Whatever was true as of its training cutoff | Unknown to the reader unless they ask |
The first three all carry a date attached to the fact. The fourth doesn’t, which is the actual problem: not that the model is wrong more often than a human researcher, but that its answer looks exactly as confident when it’s six months stale as when it’s current.
The check that catches it
Don’t trust a named seat without a dated source behind it. In practice, that’s one habit: before a seat name goes into anything client-facing, re-fetch the primary source the same day, an EDGAR search, a company’s own newsroom, a press release, rather than relying on what a model already “knows.”
For a market map, that check runs on every named seat, not just the one in the news. What’s actually in a Market Map walks through which module answers which question and from which source; the org chart module runs on the same re-fetch-before-shipping rule as the comp module.
The Opportunity Monitor exists for exactly this gap: it watches a client’s own named market against the public record continuously, so a reshuffle like UPS’s shows up the day it’s filed, not the next time someone happens to ask.
Book a call: a sample on a brief of your choice, with every seat checked against a source dated today, not a training run from months ago.
Sources
- UPS, "UPS Announces Executive Leadership Changes and New Global Operating Model Effective September 1, 2026", Form 8-K Exhibit 99.1, filed August 31, 2026.
- U.S. Securities and Exchange Commission, EDGAR filing index for the UPS 8-K, filed August 31, 2026.
- Cluen, most recent Top Trends in Executive Search Technology survey, as reported by Hunt Scanlon Media (May 2026).
Frequently asked questions
Does ChatGPT know about recent executive changes?
Only up to its training cutoff, and it typically won't tell you that unprompted. Ask a general-purpose model who holds a specific executive role and it will answer confidently from whatever it last learned, with no built-in signal for "this may have changed since then." A change announced this week, like UPS's August 31, 2026 leadership reshuffle, is invisible to every model trained before it happened.
How do you verify who currently holds an executive role?
Check a dated, primary source, not a chat answer. A company's own newsroom or investor relations page shows the announcement itself, usually the same day. An SEC 8-K under Item 5.02 is the filed, legally required version, due within four business days of the change.
Both beat a model's training data because both carry a date you can check against today.
What is a Form 8-K Item 5.02 filing?
The SEC form a public company must file within four business days of an executive appointment, departure, or role change. It names the person, the title, and the effective date, filed to EDGAR and searchable for free. It's the fastest primary source for "who holds this seat as of today" on any public company.