Blog
regulatory change managementregulatory change management processtrack regulatory changes

Regulatory Change Management: How to Track New Rules with AI

A step-by-step process for regulatory change management: find new rules, triage them, assess impact, assign owners, and use AI prompts to speed up each step.

Portrait of Luis Chavez-Mattos
Edited by Luis Chavez-Mattos, Director, Product··RSS

For Legal & Compliance, Executives and Managers

A freckled woman in a knit sweater leans toward a glowing off-frame screen, with a stack of empty tiered trays lit beside her.

What is regulatory change management?

Regulatory change management is the process of finding new or amended laws, regulations and regulator guidance, deciding whether they apply to your company, and making sure the right people change policies, controls and training before the deadline. It has five parts: monitor, triage, assess impact, implement, and keep evidence. AI helps most with monitoring, summarizing and first-pass impact analysis. A person who owns the obligation still makes every decision.

TL;DR

  • Regulatory change management is a loop, not a project. You scan for changes, triage them, assess impact, assign owners, implement, and record proof, then repeat on a fixed schedule.
  • Start with an obligations register. Without a list of what you are already required to do, you cannot tell whether a new rule changes anything.
  • AI is good at reading and sorting, weak at being a source of truth. Use it to summarize, classify and draft impact notes. Never rely on it for effective dates, citations or jurisdiction details without checking the primary text.
  • Triage is where you save the most time. Most alerts do not apply to you. A fast, consistent "does this matter?" step keeps the team from drowning.
  • Every change needs one named owner and one date. Shared ownership is how deadlines get missed.
  • Keep the evidence as you go. Auditors and regulators want to see what you decided, who decided it, and when.
  • Managers and executives need a short view, not a feed. A one-page status of open changes, owners and deadlines beats a stack of alerts.

How does regulatory change management work, step by step?

Here is the full process. Each step below has what to do, what AI can do for you, and what to check.

StepOutputWhere AI helps
1. Build an obligations registerA list of what you must comply withDrafting and cleanup
2. Set up your monitoring sourcesA defined list of feeds and alertsSummarizing and de-duplicating
3. Triage every itemApply / Watch / Ignore decisionClassification
4. Assess impactA short impact note per changeFirst-draft gap analysis
5. Assign owners and datesAn action planDrafting tasks
6. Implement and communicateUpdated policies, controls, trainingDrafting redlines and comms
7. Record evidenceAn audit trailSummarizing decisions
8. Review on a scheduleA recurring cycleTrend summaries

Step 1: How do you build an obligations register?

An obligations register is a table of the rules your company must follow, mapped to the internal policy, control or process that meets each one. It is the thing every new rule gets compared against. If you already have one, check when it was last updated. If you do not, start small.

Use these columns:

ColumnExample content
Obligation IDOB-014
Source (law, regulation, guidance)Name and section of the rule
JurisdictionWhere it applies
Plain-English requirementOne or two sentences
Business area affectedHR, Finance, Marketing, IT
Internal policy or controlPolicy name or control ID
OwnerOne named person
Last reviewedDate

Begin with the one or two regimes that carry the most risk for your business, such as data privacy, financial reporting, employment, product safety or anti-corruption. Do not try to catalog everything in week one.

AI prompt for drafting register entries. Paste in the text of a regulation section you have the right to use (not confidential material):

You are helping a compliance team build an obligations register.
Below is the text of a regulation. Extract each distinct obligation
as a row with these fields: requirement in plain English (max 2
sentences), who it applies to, any deadline or frequency stated in
the text, and the exact section reference. Quote the section
number exactly as written. If something is unclear or depends on
definitions elsewhere, write "NEEDS HUMAN REVIEW" instead of guessing.

TEXT:
[paste text here]

What to check: every section reference against the original, and every "applies to" statement against the definitions in the law. AI often smooths over a definition that decides whether you are in scope.

Step 2: Where should you monitor for new rules?

Pick a short, fixed list of sources and check them on a schedule. Mixing too many feeds creates noise. Good sources fall into four groups:

  • Primary sources: the official publications and websites of the regulators that oversee you, and the official journal or register where laws are published in each jurisdiction.
  • Industry bodies: trade associations and professional bodies, which often summarize changes in your sector.
  • Law firm and advisory alerts: useful for interpretation, but treat them as commentary, not the rule itself.
  • Internal signals: questions from the business, new products, new markets, and new vendors. A change in what you do can create new obligations even when the law has not changed.

Assign each source a check frequency (daily, weekly, monthly) and a person responsible for checking it. For high-volume sources, route alerts into one shared inbox or folder so nothing lives in one person's email.

AI prompt for a weekly digest. Paste the week's alerts and headlines into a single message:

Below are this week's regulatory alerts and articles. For each item
give me: a one-line summary, the regulator or source, the topic,
whether it is a final rule, proposal, consultation, or commentary,
and any dates mentioned (state exactly how the date is described,
for example "effective" or "comments due"). Group duplicates
together. Do not add information that is not in the text. If the
status is unclear, say so.

ITEMS:
[paste here]

What to check: the status label. Mistaking a proposal for a final rule is the most common and most costly summarizing error, so confirm status on the regulator's own page before anyone acts.

Step 3: How do you triage a new regulatory change?

Triage means making one fast decision per item: Apply, Watch or Ignore. It should take minutes, not hours.

  • Apply: the rule covers your business activities, products or locations. Move to impact assessment.
  • Watch: it is a proposal, consultation, or a rule that may apply later. Set a date to recheck.
  • Ignore: it does not touch you. Record why in one line, so the decision is defensible later.

Use the same questions every time:

  1. Is this binding (law, regulation) or advisory (guidance, opinion)?
  2. Does it apply to our jurisdictions?
  3. Does it apply to our activities, products or data?
  4. What is the effective date, and is there a transition period?
  5. Does it change something already in the obligations register, or add something new?

AI prompt for triage. Give the model your context once, then paste each item:

Context about our company: [2 to 4 sentences: industry, countries
we operate in, main products or services, types of data we hold,
number of employees range].

Here is a regulatory development. Using only the text provided,
answer:
1. Binding or non-binding, and its status (final, proposed, etc.)
2. Could it apply to a company like ours? Answer Likely / Possibly /
   Unlikely, and give the reason in two sentences.
3. Dates mentioned, quoted exactly.
4. What I would need to confirm before deciding.
Recommend Apply, Watch or Ignore, and label it as a suggestion
for a human to confirm.

TEXT:
[paste here]

What to check: the model only knows what you told it about your company. If your context is vague, the "Unlikely" answers will be unreliable. Have a compliance or legal person confirm every Ignore on anything high-risk.

Step 4: How do you assess the impact of a regulatory change?

For each item marked Apply, write a short impact note. It should answer what changes, who is affected, what must be done, by when, and how big the effort is.

Use this template:

FieldWhat to write
Change summaryTwo sentences, plain English
Source and sectionExact citation and link to the primary text
Effective date and transitionAs stated in the source
Obligations affectedIDs from your register
GapWhat we do today versus what is required
Business areas affectedTeams, systems, vendors, customers
Actions neededBulleted list
Effort and riskLow / Medium / High, with a reason
Decision ownerOne name

AI prompt for a first-draft gap analysis. Paste the relevant rule text and the matching internal policy text:

You are assisting a compliance analyst. I am giving you (A) the
text of a new or amended requirement and (B) our current internal
policy on the same topic. Compare them.

Produce a table with columns: Requirement (quote or closely
paraphrase, with section reference), What our policy says today,
Gap (None / Partial / Missing), Suggested change to our policy.
After the table, list questions a lawyer should answer before we
finalize. Do not assume facts about our operations that are not in
the text. Flag anything ambiguous.

(A) REGULATION TEXT:
[paste]

(B) OUR POLICY:
[paste]

What to check: whether the model missed a requirement (read the original yourself and tick them off), and whether "Suggested change" language is legally precise. Treat suggested policy wording as a draft for legal review.

Step 5: Who should own each change, and how do you set deadlines?

Every change that reaches the action stage needs one owner, one due date, and one definition of done. The owner is usually the head of the business area that runs the affected process, not the compliance team. Compliance coordinates and challenges. The business implements.

Work backward from the effective date. A practical plan has these dates:

  • Impact note approved
  • Policy or control changes drafted
  • Systems or process changes complete
  • Training or communication delivered
  • Evidence filed and change closed

Leave a buffer before the legal deadline. Regulators rarely accept "we were nearly ready."

AI prompt for an action plan:

Turn this approved impact note into an action plan. For each action
give: a task description, the function that should own it (use
only these functions: [list yours]), a suggested due date counting
back from the effective date of [date], and what evidence would
show it is done. Keep tasks specific and under 20 words each.

IMPACT NOTE:
[paste]

What to check: owners and dates are proposals. A person has to agree to them. A plan nobody accepted is not a plan.

Step 6: How do you implement and communicate the change?

Implementation is the part AI cannot do for you, but it can speed up the writing. Typical work includes updating a policy, changing a control or form, adjusting a system setting, updating vendor contracts, and training staff.

For communications, tailor the message to who is reading:

  • Front-line staff: what changes in their daily work, from what date, and who to ask.
  • Managers: what they must do, what they must confirm, and what the deadline is.
  • Executives: the business risk, the cost or effort, and the decision needed.

AI prompt for audience-specific messages:

Using the approved action plan below, write three short messages:
(1) for front-line employees, in plain language, under 120 words;
(2) for managers, listing what they must do and by when, under 150
words; (3) for the executive team, covering risk, effort and any
decision needed, under 100 words. Do not add requirements that are
not in the plan.

ACTION PLAN:
[paste]

Have the policy owner approve the final wording before anything goes out.

Step 7: What evidence should you keep?

Keep a record that shows the change was seen, assessed and acted on. For each item, store:

  • The date you identified it and the source
  • The triage decision and who made it
  • The approved impact note
  • The action plan with owners and completion dates
  • Updated policies (with version history), training records and sign-offs
  • The reason for any "Ignore" or "Watch" decision

This is what an internal auditor or regulator asks for. It also protects you when someone leaves and a successor asks why a decision was made.

How should managers and executives track regulatory change?

If you are a manager or executive, you do not need every alert. You need a short status that answers four questions:

  1. What new rules apply to us right now?
  2. Who owns each one, and are they on track?
  3. What is due in the next 90 days?
  4. What is overdue or at risk, and what decision do you need from me?

A one-page view with columns for change, owner, due date, status (not started, in progress, complete, overdue) and risk level covers it. Ask for it monthly, or more often when a major rule is landing. If the report is longer than a page, it is hiding the point.

For leadership meetings, ask compliance to bring: the top three changes by risk, any missed or at-risk deadlines, and any resource requests.

Where should you not trust AI in regulatory change management?

AI makes the work faster, but it fails in specific, predictable ways. Know them.

  • Invented or wrong citations. Models can produce a plausible-looking section number or case name that does not exist. Always click through to the primary source.
  • Wrong effective dates. Dates are easy to misread, especially transition periods and staggered deadlines. Verify each against the official text.
  • Proposal versus final rule. Models blur the difference. Confirm status on the regulator's site.
  • Jurisdiction and scope nuance. Whether a rule applies often depends on defined terms, thresholds and exemptions. A model may skip these.
  • Outdated knowledge. A model may not know about recent amendments unless you give it the current text. Paste the source in rather than asking from memory.
  • Confidential data. Do not paste privileged legal advice, personal data or confidential contract terms into a tool your company has not approved. Check your AI use policy first.
  • Legal advice. AI output is a draft for a qualified person to review. It is not a legal opinion.

A safe working rule: AI reads and drafts, a person verifies against the primary source, and a named owner decides.

How often should you review your regulatory change process?

Run the monitoring loop weekly for high-volume areas and monthly for slower ones. Review the process itself at least once a year:

  • Are the sources still the right ones?
  • How many alerts were Apply, Watch and Ignore? If almost everything is Apply, your triage is too loose. If nothing is, check that your context is current.
  • Did any change reach us late? Find out why.
  • Is the obligations register up to date after reorganizations, new products and new markets?

Also revisit your AI prompts. Keep a short shared document of the prompts that work, the company context paragraph, and known failure cases, so the whole team gets consistent results.

Could you build your own regulatory change tracker?

Most teams run this process in spreadsheets and shared inboxes, and it breaks down when owners, deadlines and evidence live in different places. With Remy, you can describe the tool in plain English and have it build a working app: an intake form for new changes, a register with triage status, owner and due date fields, reminder emails when deadlines near, role-based sign-in so only the right people can edit, and a one-page dashboard for managers and executives. Remy can also add AI steps, such as drafting an impact note from pasted text. You own the code, and you review the logic before relying on it.

Frequently Asked Questions

5 questions

What is the difference between regulatory change management and compliance monitoring?

Compliance monitoring checks whether you are following the rules you already have. Regulatory change management deals with the rules that are new or changing, and makes sure your policies and controls are updated to match. They connect: a new rule gets added to your obligations register, and monitoring then checks that you follow it.

Can AI replace a compliance team for tracking new regulations?

No. AI speeds up reading, sorting and drafting, but it can misstate dates, status and scope, and it cannot take accountability for a decision. Compliance and legal professionals still need to verify against primary sources and own the outcome.

How do small teams handle regulatory change management without special software?

Start with a shared spreadsheet for the obligations register and a second tab for open changes with owner, due date and status. Set a regular review schedule, use a short list of official sources, and use AI prompts for summaries and first-draft impact notes. Move to a purpose-built tool when volume or audit demands outgrow the spreadsheet.

How long does it take to respond to a new regulation?

It depends on the size of the change and the transition period. A narrow guidance update may need only a policy edit, while a major new law can take many months of work across several teams. Work backward from the effective date, and start the impact assessment as soon as you mark an item Apply.

What should I do first if we have no process today?

Pick the one or two regimes that carry the most risk and write down the obligations in them. Then choose a short, fixed list of official sources to check on a schedule, name one owner for each obligation, and start a simple log of changes with triage decisions. You can add AI summaries and impact templates once that foundation exists.

How we source, write, and correct our articles. Editorial standards

Related articles

Straight to your inbox.
New pieces as they publish. No noise.