GH-300 - Understand GitHub Copilot data and architecture - Section 3.1

Explain how GitHub Copilot handles data, including usage, flow, sharing, and plan-level differences.

Explain how GitHub Copilot uses, transmits, and shares data: how a prompt is assembled from the editor context, what is and is not retained, and how training-data use differs between the individual plans and Copilot Business or Enterprise. Recognise that Business and Enterprise prompts and suggestions are not used to train the models.

Data flowPrompt buildingData retentionTraining data

Practice question for this objective

Free sampleUnderstand GitHub Copilot data and architecturemedium

A compliance lead at a company on Copilot Business asks whether the prompts and suggestions generated by their developers are used to train the underlying foundation models. What is the accurate position for Copilot Business and Copilot Enterprise?

  • APrompts and suggestions are used for training by default, but each developer may file a support ticket to remove their data from the next training run.
  • BPrompts and suggestions are not used to train the foundation models, which is a data-handling commitment of the Copilot Business and Copilot Enterprise tiers. Correct
  • CPrompts are used for training while suggestions are excluded, because only the developer's input text carries enough signal to improve the models.
  • DPrompts and suggestions are used for training only when content exclusions are configured, since exclusions redirect the data into the training pipeline.
State that Copilot Business and Enterprise do not use developer prompts or suggestions to train the foundation models. Under Copilot Business and Copilot Enterprise, GitHub does not use developer prompts or suggestions to train the foundation models; this is a default data-handling commitment, not an opt-out, and content exclusions are a separate context control unrelated to training.

Why A is wrong: The business tiers do not train on prompts by default with an opt-out by ticket; training is excluded outright, so this invents a retraction process that does not apply.

Why B is correct: GitHub commits that under Copilot Business and Copilot Enterprise prompts and suggestions are not used to train the foundation models, which is exactly the assurance the compliance lead is seeking.

Why C is wrong: Neither prompts nor suggestions are used for training on the business tiers, so splitting them and training on prompts alone misrepresents the guarantee that covers both.

Why D is wrong: Content exclusions stop named files being used as context; they do not feed data into training, so tying training to exclusions confuses two unrelated controls.

See more GH-300 practice questions, answers explained.

More in this domain

Back to all Understand GitHub Copilot data and architecture objectives, or the GH-300 cert hub.

Examworthy is not affiliated with or endorsed by GitHub. Original, blueprint-aligned practice material only.