Sometimes. Whether Anthropic may use Claude Code code and prompts to improve future models depends on your account type and settings. Free, Pro, and Max users can allow model improvement; when enabled, Anthropic says it trains new models using data from those accounts, including Claude Code sessions. Team, Enterprise, API, and covered third-party commercial routes are not used for generative-model training by default, unless the customer opts in.
These are Anthropic’s policies as described in its documentation accessed October 2026; account terms and policy pages can change. The current Claude Code data-usage documentation is the most direct reference.
![Claude AI for Beginners Bible: [5 in 1] The Ultimate Guide to Automate Your Work, Save Hours Every Week, and Use AI for Real-World Results](https://m.media-amazon.com/images/I/415+fSJacsL._SL160_.jpg)
As an Amazon Associate I earn from qualifying purchases.
Does Anthropic train on Claude Code code and prompts?
For consumer accounts, the answer turns on the Model Improvement setting. Anthropic says data from Free, Pro, and Max accounts may be used to improve future models when that setting is on, including data from Claude Code sessions. Its current documentation describes a choice but does not state one universal default for every consumer account.
The Tool Desk
Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →For commercial accounts and routes, Anthropic says it does not train generative models using code or prompts sent to Claude Code under commercial terms unless the customer chooses to provide data for model improvement. This is a training policy, not a promise that prompts never leave your device: Claude Code sends prompts and model outputs over the network to the selected provider.
#1 Best Overall
How the policy differs by plan and provider
| Account or route | Use of code and prompts for model training | Important qualification |
|---|---|---|
| Free, Pro, and Max | May be used for model improvement when the setting is enabled. | Incognito chats are excluded from model improvement. Free is covered by the consumer policy, but the current setup guide says the Free plan does not include Claude Code access. |
| Team and Enterprise | Not used to train generative models by default under commercial terms. | Customers can opt in to model improvement; organization owners also have feedback controls. |
| Anthropic API | Not used for training by default under commercial terms. | Explicit opt-in, including the Development Partner Program, can permit use. That program is limited to Anthropic’s first-party API. |
| Bedrock, Google Cloud Agent Platform, Microsoft Foundry, and other third-party routes | Commercial default is no training on Claude Code code or prompts unless the customer opts in to model improvement. | Network, encryption, telemetry, and feedback behavior vary by provider. The Development Partner Program is unavailable to Bedrock and Google Cloud Agent Platform users. |
Sources: Anthropic’s Claude Code data-usage documentation, setup guide, and commercial training article.
How to check the consumer Model Improvement setting
For consumer Claude accounts, use Anthropic’s privacy-setting instructions to locate and change Model Improvement. The label and available controls can vary with account context, so follow the current account-specific directions rather than assuming a default.
- Open Anthropic’s Model Improvement privacy settings instructions.
- Follow the instructions for your account and check whether Model Improvement is enabled.
- If you turn it off, Anthropic says future training stops for new and previously stored chats. Data already in training runs or models already trained is not removed.
Anthropic’s consumer explanation is available in its model-training article. Incognito chats are a separate exception: Anthropic says they are not used to improve Claude even when Model Improvement is enabled.
Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minutePC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Training, retention, and local session files are different
A no-training setting does not itself answer how long information is retained, and retention does not mean that data is used to train a model. Anthropic’s current Claude Code documentation lists these periods:
- Consumer data with Model Improvement allowed: five-year retention. Anthropic’s Privacy Center also says data used for model improvement may be retained in de-identified form for up to five years in model-training pipelines.
- Consumer data when Model Improvement is off: 30 days, according to the Claude Code documentation.
- Commercial data: 30-day standard retention. Zero data retention is available for eligible Claude for Enterprise accounts but is not automatically included in standard Enterprise.
- Local session cache: Claude Code stores plaintext session transcripts under
~/.claude/projects/for 30 days by default to support resuming sessions.
These periods describe different forms of retention and are policy terms, not guarantees that code stays only on the local machine. See Anthropic’s Claude Code data-usage documentation and consumer retention article.
Feedback can send code separately from model training
Some feedback actions can transmit conversation history, including code. Anthropic says submissions through /feedback, /bug, and /share can include that history and are retained for five years.
A session-quality rating alone records a rating. If you receive a separate prompt asking whether Anthropic may inspect the transcript, selecting Yes uploads the conversation transcript, subagent transcripts, and raw session log. Anthropic says known API-key and token patterns are redacted, but source code and file content are uploaded as-is; shared transcripts are retained for up to six months. Do not treat a rating by itself as transcript sharing.
What “not used for training” does and does not mean
Claude Code runs locally, but sends prompts to the selected model provider and receives model outputs over the network. Operational metrics are a separate stream; Anthropic’s documentation says they exclude code, prompts, and file paths. Provider-specific encryption-at-rest, telemetry, and network practices differ, so a plan label alone cannot establish every detail of a deployment. Check the applicable provider terms and your organization’s agreement for stricter handling requirements.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
