Enterprise FAQ

Data and privacy

Is my code stored by Anthropic?

No, not by default. With the direct API, retention is configurable (0 to 30 days). With AWS Bedrock or Google Vertex AI, no data passes through Anthropic's servers. Code sent during a session is only used to generate the response; it is not used to train models.

What data is sent exactly?

Claude Code sends your prompt, files it reads during the session, and results of executed commands. It does not automatically send your entire source code. .env files, SSH keys, and system credentials are never sent unless you explicitly ask Claude to read them.

Does Anthropic use my data to train its models?

No. Data sent via the API is not used for model training. This is explicitly stated in the API terms of service and in the DPA.

Can Claude Code be used on a disconnected internal network?

Not directly. Claude Code requires a connection to Anthropic's API (or a Bedrock/Vertex endpoint). For air-gapped environments, the only option is an internal proxy that routes requests to the API through a controlled exit point.

Compliance

Is Claude Code GDPR-compliant?

Anthropic offers a GDPR-compliant DPA (Data Processing Agreement), with standard contractual clauses for EU-US transfers. To strengthen compliance, use AWS Bedrock in a European region (eu-west-1, eu-central-1). Consult your DPO for an analysis tailored to your context.

What about the European AI Act?

Claude Code, used as a development assistance tool, falls into the "limited risk" category of the AI Act. The main obligations are transparency (informing users they are interacting with an AI) and usage documentation. If the generated code feeds into a high-risk system, additional obligations apply to the final system.

What certifications does Anthropic hold?

SOC 2 Type II (obtained), HIPAA (BAA available on request), standard GDPR DPA. ISO 27001 is in progress. Audit reports are available on the Anthropic Trust Center.

Can we get a security audit from Anthropic?

SOC 2 reports are available on request through the Trust Center. For additional audits or penetration testing, contact Anthropic's sales team.

Costs and billing

How much does Claude Code cost for a team of 50 developers?

With the Max 5x plan ($100/dev/month): $5,000/month. With the API (average consumption of $40/dev/month): about $2,000/month. The actual cost depends on usage volume. Use the TCO calculator for a tailored estimate.

Is the $20 Pro plan enough for a team?

For occasional use (a few requests per day), yes. For intensive daily use, throttling on the Pro plan slows down the workflow. Most teams that use Claude Code seriously move to the Max or API plan.

Can plans be mixed within a team?

Yes, nothing prevents it. You can have devs on Max 5x (intensive use) and others on Pro or API (moderate use). That's actually the recommended strategy for optimizing costs.

How do you track consumption on the API plan?

Via the Anthropic dashboard (console.anthropic.com) or through your cloud provider's metrics (Bedrock/Vertex). You can also set budget alerts to be notified before reaching a threshold.

Deployment and training

How long does it take to deploy Claude Code in a team?

Plan 1 week of preparation, 4 to 6 weeks of pilot, then 2 to 4 weeks for progressive rollout. Total: 2 to 3 months for solid adoption. The adoption guide details each phase.

Is specific training needed?

A half-day initial training is enough for the basics. The most important thing is to train on effective prompting, not installation (which takes 5 minutes). Also plan weekly follow-ups during the first weeks.

Does Claude Code work with all languages?

Yes. Claude Code understands and generates code in all popular languages: TypeScript, Python, Rust, Go, Java, C#, Ruby, PHP, Swift, Kotlin, and many others. It performs especially well on TypeScript, Python, and Rust thanks to the volume of training data.

Can Claude Code be used with a monorepo?

Yes. The CLAUDE.md file at the monorepo root defines global rules, and each package can have its own CLAUDE.md with specific instructions. The 200K token context window supports working on large projects, but use /compact regularly to avoid saturating the context.

Intellectual property

Who owns code generated by Claude Code?

Code generated by Claude Code belongs to you. Anthropic's terms of service state that outputs generated via the API are the property of the user. Anthropic claims no rights over generated code.

Is generated code subject to copyright?

The question of copyright for AI-generated code is still being debated legally in several countries. In practice, code generated by Claude Code is similar to code written with the help of a tool: you can modify, distribute, and commercialize it freely. Consult a lawyer if your situation requires it.

Is there a risk of contamination from open source code?

Claude Code generates original code based on its knowledge, not copy-pasted from existing code. The risk of literal reproduction of licensed code is low but not zero, especially for very common patterns. For sensitive projects, run generated code through a license detection tool (like FOSSA or Snyk).

Support and reversibility

What support does Anthropic offer for enterprises?

Standard support is available through the Anthropic website. For enterprise teams, priority support is available with dedicated SLAs. Contact the sales team for details.

What happens if we stop using Claude Code?

Nothing blocking. Claude Code modifies your local files, like any other tool. Your code stays your code, in your git repo. There is no technical lock-in: no proprietary format, no runtime dependency. You just lose the assistant's help.

Can you migrate from Claude Code to another AI tool?

Yes. The configurations (CLAUDE.md, settings.json) are specific to Claude Code, but Skills and prompting best practices are largely transferable. MCPs are based on an open protocol, and some are compatible with other tools.

Integration and workflow

Does Claude Code integrate with our CI/CD?

Yes. Headless mode (claude --print) lets you integrate Claude Code into GitHub Actions, GitLab CI, or any pipeline. See the headless and CI/CD guide for details.

Can Claude Code be used with VS Code or JetBrains?

Claude Code runs in the terminal, independently of the editor. It combines naturally with any IDE: you keep your favorite editor and use Claude Code in a terminal alongside it.

Impact on team dynamics: will juniors stop learning?

This is a valid concern. In practice, juniors who use Claude Code learn faster because they have a "mentor" available 24/7 that explains code, patterns, and errors. The key is to encourage them to understand generated code, not blindly copy it. Code reviews remain essential.

Next steps

• Security and compliance: details on data protection
• TCO calculator: estimate the budget for your team
• Adoption guide: launch your structured pilot