Workers Risk Dismissal for Using AI Wrongly as Hidden Data Can Still Expose Identities
Updated
Updated · Computerworld · Nov 16
Workers Risk Dismissal for Using AI Wrongly as Hidden Data Can Still Expose Identities
3 articles · Updated · Computerworld · Nov 16
Summary
Employees using generative AI at work can lose their jobs even without sharing obvious secrets, because using an unapproved tool may itself violate company rules.
Hidden identifiers make that risk hard to judge: error logs, file paths, screenshots, hidden spreadsheet columns and meeting transcripts can reveal proprietary systems or re-identify people after names are removed.
One therapist was dismissed after using ChatGPT instead of an approved AI tool to build a billing spreadsheet framework, despite saying she entered no confidential client data.
Security and legal advisers say employers should set boundaries before staff use AI—covering approved accounts, permitted tasks, contracts, retention and access controls—then reinforce them with examples, training and technical safeguards.
For workers under deadline, the practical fallback is slower: recreate problems with fictional data, use internal resources when context is too sensitive, and get written guidance before using outside AI services.