AI Threat Watch

Case mchire-paradox-exposureModel & data leakUnited States

McDonald's McHire AI hiring bot exposed up to 64M applicant records via '123456' login

Researchers accessed the admin panel of the Olivia hiring chatbot with default 123456 credentials and an IDOR flaw, exposing chat logs and contact data tied to up to 64M applications.

HighAttack on AIStatus: PatchedVerified case file

Open in the interactive mapCite this case

Reported
Jul 9, 2025
Location
United States
Target
McDonald's / Paradox.ai
Severity
High
Triggered by
Security researchers Ian Carroll and Sam Curry (responsible disclosure)
Actor type
Security researchers
Reported loss
None reported
CVE
None

What happened

Researchers accessed the admin panel of the Olivia hiring chatbot with default 123456 credentials and an IDOR flaw, exposing chat logs and contact data tied to up to 64M applications.

Impact: Names, emails, phone numbers, addresses and chatbot transcripts tied to up to 64 million McHire applications were reachable; Paradox said only the researchers accessed a portion of records

Who triggered it

Actor: Security researchers Ian Carroll and Sam Curry (responsible disclosure)

Attribution: Self-disclosed by the researchers; confirmed by Paradox.ai and McDonald's

Motive: Security research; no malicious access reported

How it happened

Attack vector: Default credentials 123456/123456 on a Paradox.ai test franchise account in the McHire admin portal, plus an IDOR on an internal API allowing lead_id enumeration

Root cause: Unchanged default test credentials on a production-connected admin portal and missing authorisation checks on an applicant-data API

Timeline

  1. Researchers find the flaws and report to Paradox.ai and McDonald's
  2. Default credentials disabled and IDOR fixed the same day, per McDonald's and Paradox.ai
  3. Public disclosure via Wired and BleepingComputer; Paradox.ai announces a bug bounty programme

Response

Paradox.ai and McDonald's fixed the issues the day they were reported; Paradox launched a vulnerability disclosure bounty programme

Current status: Patched

Sources (open in a new tab)

Cite this case

McDonald's McHire AI hiring bot exposed up to 64M applicant records via '123456' login. AI Threat Watch, case mchire-paradox-exposure, Jul 9, 2025. https://aithreat.io/cases/mchire-paradox-exposure/

Permanent link: https://aithreat.io/cases/mchire-paradox-exposure/

This case file in other languages: EnglishFrançaisEspañolالعربية中文