Somebody in your organisation has already asked whether the SOPs need updating. It is a reasonable question and the answer is no — but the reasoning matters, because the same question will come back every time vocabulary shifts.
The US executive order of 29 September 2026 tells federal agencies what words to use in their own non-statutory materials. It says nothing about what a private company calls anything in its quality system.
Why a find-and-replace is the wrong reflex
Renaming AI to SI across your document set would cost you, in exchange for nothing:
- A change control per document, each needing assessment and approval.
- A retraining obligation. Revise an SOP and the people who follow it generally need read-and-understood records against the new version.
- A version history full of cosmetic revisions, which makes real changes harder to find later.
- Inconsistency with everything else — your EU documentation, your vendor contracts, your GAMP references and your existing validation packages would all still say AI.
And you would be doing it for a term that is, by the order's own definition, a synonym.
The one change worth making
Add a single sentence to your AI governance policy or validation procedure, at its next scheduled revision rather than as an emergency change:
"US federal materials may use the terms 'Super Intelligence' and 'SI' for what this procedure calls artificial intelligence. The terms are equivalent for the purposes of this procedure."
That sentence does the whole job. It means a reader who encounters SI in an FDA document knows the procedure covers it, and it means an inspector asking about SI gets a document that plainly addresses the question.
Writing documents that survive vocabulary changes
The deeper lesson is worth taking, because this will not be the last relabelling.
- Define your terms once, centrally. If "artificial intelligence" is defined in one policy and referenced everywhere else, a future change is one document, not forty.
- Describe behaviour, not branding. A procedure that says "systems whose output is not fully determined by explicit rules" keeps working regardless of the label. One that says "AI tools" depends on a word staying in fashion.
- Scope by risk and function, not by technology name. Your controls should attach to what the system decides and what happens if it is wrong — the basis of any sound GxP risk assessment.
If your document set would need forty changes to absorb one new synonym, that is worth knowing, and it is a finding about your documentation architecture rather than about this order.
Background in what the rename changes for GxP, and the training implications in updating SOPs and training for AI.
Where to go next
Explore GxP Copilot for AI-native validation, TraceDraft for source-traceable clinical documentation, or book a demo to see either on your own data.