Confirm every named person's current title and company directly on a primary source, such as the company's own leadership page, an official bio, or a recent first-party announcement, rather than on a secondhand summary or an older article the AI may have drawn from. Titles change constantly through promotions, departures, and acquisitions, and a language model's knowledge of any given person is frozen at whatever point its training data stopped, which is often well behind the present. A name and title that were correct at some point are not automatically correct now, and that gap is exactly where this kind of error lives.
Why a wrong name or title is the error readers notice first
A reader who has never checked a statistic in their life will still notice, immediately, if you get a CEO's name wrong or credit someone with a job they left a year ago. These are the errors that make a piece look sloppy even to someone with no way to check the more technical claims, because a name is something readers frequently already know. That makes this category of error disproportionately damaging to credibility relative to how easy it is to check, which is the whole argument for treating it as a fixed step rather than something caught only if it "looks wrong."
Step 1: Confirm the spelling before anything else
Search the full name as written and see whether it matches how that person or company actually spells it. AI-written drafts occasionally produce a name that is close but not exact, a transposed letter, a common misspelling, or a near-miss with someone else who has a similar name. This is the fastest check to run and the easiest one to skip because a near-correct name does not visually flag itself as wrong the way a broken link does.
Step 2: Check the current title on the company's own page, not an aggregator
Go to the person's official bio, the company's own leadership or team page, or a recent first-party press release, rather than a business directory, a news aggregator, or an older article that may itself be out of date. Company "About" and leadership pages are usually kept current because they are the organization's own public statement about who holds what role, which makes them a better primary source for this specific check than almost anything else available.
Step 3: Watch for people who have since left, been promoted, or changed companies
A title that was accurate at some earlier point stops being accurate the moment that person moves on, and an AI model has no way to know that a departure or promotion happened after its training data ended. This risk is highest for fast-moving roles like marketing and product leadership, and for smaller or newer companies where leadership changes are less consistently covered by outside press. If the AI's draft names someone in a role at a young or fast-growing company, treat the title as unverified until you confirm it against a current source.
Step 4: Watch for companies that have been renamed, acquired, or merged
The same staleness problem applies to company names. A company mentioned by an older name, a name it used before an acquisition or rebrand, or a subsidiary now folded into a parent brand, can all appear in an AI draft as if the older name were still current. Check the company's own site directly for its current legal or brand name, and note if the entity described no longer exists in the form the draft implies.
Step 5: Check for people who share a name
A name search can surface the wrong person entirely, especially for common names or when the AI has merged details from two different people into one description. Cross-check identifying details beyond the name alone, such as the company, the specific role, and any other biographical detail the draft attaches, against the actual person's own public profile, before assuming a name match is the right person.
The gap most guidance on this misses: writing an honest "as of" date
Almost nothing written about verifying AI content addresses what to do once you have confirmed a title is current today. The honest answer is that "current" has an expiration date the moment you publish. For any role or company detail that could plausibly change, attach a visible "as of [date]" note next to the claim, rather than stating it as a flat, permanent fact. This does two things a silent, undated claim cannot: it tells a future reader exactly how stale the information might be, and it protects the piece from looking wrong later for a fact that was genuinely correct when it was written.
A short verification checklist
| Check | What confirms it |
|---|---|
| Spelling | Matches how the person or company spells their own name |
| Current title | Confirmed on the company's own leadership or bio page, not a secondhand summary |
| Still employed there | No recent departure, retirement, or move you can find |
| Company name | Current legal or brand name, not a pre-acquisition or pre-rebrand name |
| Right person | Other identifying details match, for common or shared names |
| "As of" date | Attached to any detail that could plausibly change after publishing |
This pairs with why AI gets dates and recent events wrong, since a stale title is one specific case of the broader training-cutoff problem, and with how to fact-check anything an AI writes for you for where this step sits in the full process. If the concern is a quote rather than a title, see checking a quote an AI attributed to someone.
FAQ
How do I check a title if the company has no public leadership page?
Look for the person's own professional profile, a recent conference bio, or a first-party press release naming their role, and prefer whichever source is most recently dated. If nothing current is findable, state the title with a visible "as of" date tied to your best available source rather than presenting it as settled fact.
Is a news article a good enough source for someone's current title?
Only if it is recent. An article from a year or more ago is describing that person's title at the time it was written, not necessarily now, and should be treated the same as any other potentially stale source.
What if the AI got the company right but the title slightly wrong, like "VP of Marketing" instead of "SVP of Marketing"?
Treat it as an error worth fixing, not a rounding difference. Titles carry specific meaning inside an organization, and a reader or the person themselves may notice a title that is close but not accurate.
Does this matter as much for a smaller or less well-known company?
It matters more, if anything, because smaller companies are covered less by outside press, which means there is less independent verification available and a higher chance the AI's information is drawn from a single, possibly outdated, source.
What is the fastest way to check several names in one draft?
Pull every named person and company into a list first, then work through each one against a primary source before touching anything else in the piece, the same triage-first approach that works for statistics and citations generally.