You will not find “pasted into ChatGPT” in Workday’s Process Monitor. Detection is an AMS operating problem.
Signals
- Personal Free/Plus/Pro AI accounts used on project days (training defaults on for ChatGPT personal workspaces unless opted out).
- Unlabeled screenshots of worker profiles in personal cloud drives.
- Local Claude Code project folders under home directories containing client exports (
~/.claude/projects/retention defaults apply). - ISU credentials in assistant configs (compound failure with shared accounts).
Response
- Written ban on personal consumer accounts for client content.
- Firm-managed AI seats or zero-retention credits path.
- Offboarding wipes for local assistant transcripts.
- Spot checks on a sample of consultants each quarter—not surveillance theater without a policy.
Detection without a sanctioned alternative just drives pastes into private phones.