When I shared my last article, a few people got in touch asking for a more practical follow-up, specifically around how small teams can use Claude Pro without putting business data at risk.
This piece goes step by step through exactly that.
Understand what you're actually adopting
Claude Pro is Anthropic’s paid consumer subscription tier, priced at £18/month per user. It gives individuals access to Claude’s most capable models, priority access, longer context windows, and the ability to upload files and use Projects.
It is designed for individual professionals, not for teams or organisations.
That distinction matters more than most people realise.
CRITICAL: Claude Pro is a consumer product. When your employees use Claude Pro, whether with a personal email or a work one, they are operating under Anthropic’s Consumer Terms of Service, not a business agreement. This has serious implications for data privacy, training consent, and compliance. Most users don’t appreciate this before they start using it.
Your team are probably already using it
If you haven’t formally provided a Claude subscription to your team, the chances are they’re already using it, on personal accounts, free tiers, or personal Claude Pro subscriptions.
That means your business data is potentially entering a consumer AI platform with no oversight, no audit trail, and no data protection agreement in place. This is what’s known as shadow AI.
The solution isn’t to ban Claude. It’s to provide a governed, approved route that gives your people the tool they want, safely.
CloudGuard insight: In organisations without a formal AI tool, our assessments consistently find 15–30% of knowledge workers already using personal Claude or ChatGPT accounts for work tasks. You cannot govern what you cannot see.
Consumer vs business: why the distinction matters
From September 2025, Anthropic updated its consumer terms so that Claude Pro conversations may be used for model training unless users actively opt out. Many employees accepted those updated terms without realising the implications for their employer.
Here’s how the tiers compare from a Claude Pro business security standpoint.
| Plan | Legal basis | Data retention | Training on your data | Admin controls |
|---|---|---|---|---|
| Claude Pro (Consumer) | Consumer Terms, no DPA | 30 days (opt-out of training) | Opt-out required after Sep 2025 | None |
| Claude for Work / Team | Commercial Terms + DPA included | Not used for training by default | Never used for training | Admin dashboard, user management |
| Claude Enterprise | Commercial Terms + DPA + ZDR option | Zero Data Retention available | Never used for training | SSO, audit logs, SAML, full governance |
Always verify current terms at anthropic.com before purchasing.
How to choose the right account type
Option 1: Claude Pro (Consumer), proceed with understanding
If budget constraints mean Claude Pro is your only viable option right now, you can use it more safely, but you must take specific steps first.
Claude Pro does offer some stronger defaults than many competitors: conversations are not used for AI training by default, communications are encrypted in transit (TLS 1.2+) and at rest (AES-256), and Anthropic does not sell your data to third parties.
However, Claude Pro has no Data Processing Agreement, no audit logging, no admin controls, and no formal compliance documentation. For any business handling client data, financial records, or regulated information, Claude Pro alone is not sufficient, regardless of whether the subscription is personal or company-approved.
Option 2: Claude for Work / Team, the recommended starting point
For most small businesses, Claude for Work is the right entry point where any customer data will be processed. It provides the commercial protections that Claude Pro does not:
- Commercial DPA: A Data Processing Agreement is automatically included, essential for GDPR compliance
- No training on your data: Anthropic will not use your conversations to train models
- Admin controls: Central account management, user provisioning, and usage visibility
- Audit capability: Conversation logging to support incident response and compliance
- Business continuity: Centralised billing and access management, not tied to individual employee accounts
Recommendation: If you are deploying Claude for any work involving client information, financial data, personal data, or confidential business content, Claude for Work is the minimum appropriate tier. Appoint a business owner to oversee governance of usage. The risk reduction is significant.
Option 3: Claude Enterprise, for regulated sectors or larger teams
If your business operates in a regulated sector (legal, financial services, healthcare, public services, or accountancy), or you have more than 20 users, Claude Enterprise provides the full security stack:
- SSO/SAML integration with Microsoft Entra, Okta, and Google Workspace
- Zero Data Retention option: inputs and outputs not stored beyond immediate processing
- BYOK encryption (H1 2026): manage your own encryption keys for full data sovereignty
- SOC 2 Type II and ISO 27001 compliance documentation
- Custom usage policies: set organisation-specific guardrails on what Claude can and cannot do
The UK GDPR point most guides miss
Claude Enterprise is designed to support UK GDPR compliance and provides the contractual and technical frameworks to use it lawfully, but compliance is a shared responsibility. Anthropic provides the infrastructure and contractual basis. Your organisation must implement appropriate data governance on top.
The critical point on data residency:
By default, Claude Enterprise processes data in US-based infrastructure. For UK organisations, any personal data entered into Claude prompts, including employee names, client details, or any identifiable information, may be transferred to and processed in the United States. That is an international transfer under UK GDPR.
Anthropic’s DPA includes the IDTA (the UK’s equivalent of Standard Contractual Clauses) to make that transfer lawful. But it still requires you to conduct a Transfer Risk Assessment. This is not optional in the eyes of the ICO.
If you need UK or EU-only data processing, the route is via AWS Bedrock or Google Cloud Vertex AI with regional endpoints configured, not directly through claude.ai Enterprise.
Your organisation also remains responsible for conducting a DPIA, establishing a legal basis for each use case involving personal data, updating your privacy notices to disclose AI processing and international transfers, defining data classification rules (special category data should be prohibited from Claude entirely), and maintaining a ROPA entry listing Anthropic as a sub-processor.
Not sure where to start?
CloudGuard’s Securing AI service can help if you want expert eyes on your current AI posture before it becomes a problem. We work with small and medium businesses across the UK and Europe to build AI governance frameworks that are proportionate, practical, and genuinely protective.
- AI Risk Assessment: Identify your current AI exposure, including shadow AI, and get a prioritised action plan
- AI Use Policy Development: We draft a policy tailored to your business type, data environment, and regulatory obligations
- Staff Awareness Delivery: Practical, role-specific AI awareness that changes behaviour, not just awareness scores
- Ongoing AI Governance: Monthly monitoring, vendor reassessment, and policy evolution as the landscape changes
CloudGuard | Security Done Different | cloudguard.ai [email protected]
