Why AI confuses you with someone else — and how to stop it.
If an engine has ever credited you with someone else's career, you've met the entity resolution problem. It's fixable — but not by luck.
Why did ChatGPT say I wrote someone else's book?
AI systems don't look you up — they resolve you. When a model is asked about "Michael Chen," it has to decide which of the thousands of Michael Chens the question means, then assemble an answer from records it believes belong to that one person. When the signals are ambiguous, records merge. You inherit a stranger's employer, their controversy, or their book — and they inherit yours.
Who gets misidentified most?
Three groups carry the most risk: people with common names, people who share a name with anyone even modestly public, and people whose own footprint is thin — because an engine with little to retrieve about the real you will happily pad the answer with a louder namesake. The less you've published, the more of "your" answer belongs to someone else.
Fix 1: Pick one name form and never vary it
Rob Martin on LinkedIn, Robert J. Martin on your bylines, and R.J. Martin on the conference site read as three weak entities instead of one strong one. Choose a single professional name form — middle initial included if your name is common — and enforce it everywhere your name appears in text an engine can retrieve.
Fix 2: Build disambiguating anchors
Engines separate identities using co-occurring facts: name + company + role + city, appearing together, repeatedly. Make sure your profile's opening lines contain that full anchor — "Jane Alvarez, VP of Regulatory Affairs at Meridian Bio in Boston" — and that the same anchor appears in your bios, bylines, and speaker pages. Every consistent co-occurrence is a wall between you and your namesake.
Fix 3: Interlink everything you control
Your LinkedIn, personal site, company bio, and published work should reference each other. Cross-links teach retrieval systems that these records are one entity — and quietly demote the namesake's records from answers about you. If you share a name with someone prominent, a one-line differentiator ("not the novelist") on your own site is not beneath you; it's machine-readable disambiguation.
Fix 4: Watch for it coming back
Identity blends recur. A model update or a namesake's news cycle can re-merge records you spent months separating. This is exactly what drift monitoring is for — a weekly check that the engines still know which one is you, before a client or recruiter finds out they don't.
Am I being confused with someone right now?
There's one way to know. Run your free Authority Snapshot — if any engine's answer contains a fact that isn't yours, you'll see it in minutes.