AI Coding Assistant Policy
Purpose
AI coding assistants can improve speed, but generated code must be treated as untrusted until reviewed, tested and licensed appropriately.
AIC Standard
AI tools MAY be used for:
boilerplate generation
test case ideas
refactoring suggestions
documentation drafts
code explanation
migration assistance
prototype exploration
AI tools MUST NOT be used to bypass engineering judgement, security review, licensing review or code review.
Prohibited Inputs
Do not enter into unauthorised AI tools:
customer confidential information
classified information
secrets, keys or tokens
proprietary source code where tool terms do not permit it
personal data unless approved
contractual or security-sensitive details without approval
Generated Code Rules
Generated code must:
be reviewed by a competent developer
be understood by the team
comply with AIC standards
pass tests
pass static analysis
be checked for licensing risk where material
not introduce unsupported dependencies
not include fabricated APIs
not weaken security controls
Review Checklist
Do we understand every line?
Does it compile?
Does it follow architecture boundaries?
Are tests present?
Are errors handled?
Is security preserved?
Are dependencies approved?
Is there any licensing concern?
Was this helpful?

