Sourcegraph Cody
Sourcegraph says it and its partner LLMs do not use customer code to train models, and partner LLMs operate under zero-retention terms for inputs, outputs, and code context subject to a narrow abuse-prevention exception.
Customer code is not used to train Sourcegraph or partner models. A custom model may be fine-tuned only for a customer's proprietary use if that service is purchased. Individual Sourcegraph.com users may have prompts and responses collected to improve experience, but not to train generally available models.
No general training opt-out is required for customer code under the cited AI terms. Enterprise customers can bring their own LLM key, but partner assurances may not apply to that path.
Partner LLMs do not retain input, output, or candidate code context beyond generation except temporary abuse or serious-harm retention. Sourcegraph may persist inputs and outputs to enable AI-tool functionality.
Partner LLM zero retention is subject to temporary abuse retention. Sourcegraph's own retention of inputs and outputs is not quantified in the cited AI terms.
The reviewed AI terms do not provide one general human-review statement.
Foundational reference pages for teams adopting, deploying, and operating production AI with inspectable controls.