AI Leadership Exposure Report

External Identity Exposure at the Top Leading AI Companies

A study of the external identity data exposure of 75 senior leaders across nine AI and frontier-technology companies, using only open-source data AI tools.

100% of them do not have a clean footprint.

Executive Summary

AI Leadership Exposure Report by VanishID

VanishID studied the external identity data exposure of 75 senior leaders across nine AI and frontier-technology companies. We used open-source data only, the same sources available to anyone with a laptop and a few minutes. AI has enabled this data to be found and assembled faster than ever before.

The people building artificial intelligence are among the highest-value profiling targets on the open web, and the tools they ship are what make that profiling fast.

The pattern holds across every leader we measured. A personal phone number, a date of birth, and a home address, pinned to the same person, are enough to start a SIM swap, which is a fraud technique where an attacker tricks or bribes a phone carrier into transferring a victim’s phone number to a SIM card the attacker controls, or a convincing impersonation. That combination is available for 87% of this cohort. This combination can provide useful information for impersonation and social engineering. The study did not establish whether the phone numbers were mobile lines, whether addresses were current, or whether the information would be sufficient to bypass carrier protections.

Now, add a headshot on the corporate bio page and a voice sample from any recorded panel, podcast, or earnings call. Stack them on top of the trio listed above and the same dossier supports a voice clone aimed at the help desk, a deepfake on a video call, and a vishing or smishing attempt against a direct report. Add family members, decades of breach history, and plaintext passwords still in circulation, and an attacker has a running start on either a digital account takeover or a visit to someone’s front door.

What follows breaks that chain into the ten specific attacks it enables, one real case or number at a time, and ends with what to do about it.

Copyright © 2019 – 2026 Picnic Corporation (dba VanishID)