Lead people work when legal blocks AI on PII.
Lead People and HR work when legal blocks AI on PII. The barrier is the data wall, not another enablement workshop.
Counsel sent eight pages back on a Slack onboarding bot. HR data is PII. The model cannot touch it. Meanwhile someone books an enablement workshop and asks managers to champion AI on people work leadership has not committed, and legal has not allowed. The wall is the job. The workshop is a miss.
Stay on the wall. This is not a managers-as-change-layer program. This is not a person refusing an AI coach. This is not a statute memo and not a shadow-IT hunt. People still have to run HR. They have to do it without pretending the files are in play.
You cannot enable a use the memo already killed.
Eight pages for a Slack bot.
Operator pattern, not a cited sample. Across a portfolio or a single company, the shared frustration is the same: you cannot put HR data, PII, or audit reports into a model, even on an enterprise plan. A basic onboarding helper in Slack draws a long legal response. The People team is still asked to modernize. IT owns the tool budget. Counsel owns the wall (who owns AI adoption, IT or HR).
IT-led rollouts stall when the human constraint is unread (why IT-led AI rollouts stall). Here the constraint is sharper: the data class is blocked. A policy PDF is also not enablement (AI policy is not enablement). The policy may be the only honest artifact you have.
The barrier is the data, not the workshop.
PII is the off-limits set
HR files, audit reports, performance notes, anything that identifies a person. Established fact: counsel can block models on that set even inside an enterprise account. The workshop cannot unlock it.
A basic bot can draw an eight-page memo
Operator pattern: a Slack onboarding helper that looks harmless from the People seat draws a legal response that takes a week to read. That is the wall announcing itself. It is not a training gap.
Workarounds stay unresolved
Anonymize the file. Use a vendor with a DPA. Run the model only on job descriptions. Those are live debates. They are not a green light. Do not brief the company as if they were.
Category language for AI in the People stack still matters (AI for HR). So does a platform that creates adoption after training (platforms for AI adoption). Neither page is this job while the files are off limits. Do not brief a product as if the wall moved.
Not a product pitch. ADAPT is coaching-led change management for AI adoption after the work is allowed. The ADAPT page can wait. This page will not sell around counsel.
Managers cannot champion an uncommitted wall.
Operator view: managers cannot champion what leadership has not committed. That sentence is true in any AI rollout. It is decisive here. If execs have not said which people workflows are in play, and counsel has said PII is not, a champions club is theater. HR still has to sit between strategy and employee fear (how HR can lead AI transformation).
Enterprise chat rollouts that stay on non-PII work can continue (ChatGPT Enterprise rollout). That is a different data class. Do not blur it with performance files so the workshop has a story.
What this is not.
| If you are actually deciding | Go here instead |
|---|---|
| Leadership committed, managers must own team adoption | Make managers the AI change layer |
| A person will not use an AI coach | AI coaching |
| Where AI fits in the People stack at all | AI for HR |
| Which platform creates adoption after training | Platforms for AI adoption |
After the wall moves and leadership commits, use make managers the AI change layer. A refusal to use a coaching bot is AI coaching. Come back here when the memo is the blocker, not the mindset of a single user, and not a missing workshop.
What People still owns while the wall stands.
Recommendation, not a benchmark: publish a one-page wall. Allowed: non-PII drafting, public process docs, manager practice that never uploads a file. Blocked: HRIS extracts, notes, ratings, anything that names a person. Escalation: who in counsel clears a new use. Then run People work inside those lines. Do not run another enablement hour that ignores them.
What good looks like. Managers can repeat the wall in a 1:1 without guessing. Leadership has said what is committed. A new bot request has a counsel owner. If the only change is a lunch-and-learn, you decorated the memo.
Work the wall instead of booking another enablement hour.
Bring the memo and the People workflows that are stuck. We will map what is still allowed, or tell you this is already an adoption page.
Book a strategy callFrequently asked questions
What do you do when legal blocks AI on HR and people data?
Treat the wall as the job. Stop running enablement workshops that assume the data is usable. Write what is blocked, what is allowed on non-PII work, and what leadership has actually committed. Managers cannot champion a use they are not permitted to touch.
Is this an AI adoption program for managers?
No. Making managers the change layer assumes leadership already committed and the work is allowed. This page is earlier: legal has blocked AI on PII. The enablement overlay will fail because the data is off limits.
Is this about refusing an AI coach?
No. Refusing an AI coach is a person who will not use a coaching bot. This page is a counsel memo that blocks models on people files. Different wall. Different owner.
Do we need an EU AI Act memo?
Not from this page. Regulation can be a reason counsel writes the memo. The operator job here is the wall in front of you: what people data is blocked, and how People work continues. A statute explainer is a different document.