Law · GMLP

GMLP for legal AI workflows

GMLP in legal AI workflows: what compliance requires, who asks for evidence, and the records to keep.

Legal forceNon-binding regulatory guiding principles
Applies from27 Oct 2021
SectorLaw
Checked21 Oct 2026

01 / The rule

What GMLP requires

Manufacturers developing AI/machine-learning-enabled medical devices for the US, Canadian and UK markets, and the regulators reviewing them.

  1. 01Use multidisciplinary expertise throughout the total product lifecycle and follow good software engineering and security practices.
  2. 02Use clinical study participants and data sets representative of the intended patient population.
  3. 03Keep training data sets independent of test sets, and base reference datasets on best available methods.
  4. 04Tailor model design to the available data and intended use, and evaluate performance of the human-AI team.
  5. 05Test device performance under clinically relevant conditions and give users clear, essential information.
  6. 06Monitor deployed models for performance and manage retraining risks.

02 / In this sector

In a legal AI workflow

AI-specific. The principles were written for medical devices that use machine learning, including models that are retrained or updated after release.

Usually asked by

General counsel and litigation support.

Named alongside

FRE 702 · ABA Model Rules.

Sector pagePlanisphere for law

03 / Records

Records to keep

  1. 01Dataset documentation showing representativeness and train/test separation
  2. 02Reference standard (ground truth) method and rationale
  3. 03Clinical performance testing report under intended-use conditions
  4. 04Post-market performance monitoring plan and retraining change records

Optional evidence · where Planisphere fits

Keep a record anyone can check

When your team meets a duty, Planisphere can seal the event it reports as a receipt.

01 · Your teamRecord the event

A disclosure shown, a review signed, a model change approved: the event your system reports.

02 · PlanisphereSeal it

The event becomes a sealed receipt, dated and attributed to whoever is signed in.

03 · AnyoneCheck it

Paste the receipt into Verify. No account or key.

A record supports your compliance evidence process. It is not a legal finding or a certification. Systems that call Planisphere's gate API can also hold or refuse an action before it runs.

SOURCES

Go back to the source

PRIMARY
Good Machine Learning Practice for Medical Device Development: Guiding Principles

Medicines and Healthcare products Regulatory Agency (GOV.UK)

SECONDARY
IMDRF finalizes good machine learning practice

Regulatory Affairs Professionals Society

RELATED

Related: GMLP: the full guide · GMLP for defense · GMLP for medicine · GMLP for education