Workshops: AI Tools at Work

How to Use Claude Safely at Work: Workshop for Companies

This workshop shows how your company can use Anthropic's Claude productively while protecting data. You learn how consumer and business plans differ, assess connectors and file access by risk, and write binding rules for your team.

Who is this workshop for?

  • Executives and decision makers who want to approve Claude
  • Data protection and information security teams
  • IT administration and enterprise architecture
  • Teams already testing Claude on personal accounts
  • Works council and compliance officers

What you will take away

  • You understand how Free, Pro and Max differ from Team and Enterprise in data use, administration and contractual basis.
  • You can judge which categories of data may be processed in Claude.
  • You assess connectors, web search and file access by risk and define approvals.
  • You recognise typical attack paths such as prompt injection via documents, websites or connected tools.
  • You produce a draft Claude usage policy with clear responsibilities.

Agenda

  1. Module 1

    Claude and your data

    • Consumer plans vs Team and Enterprise
    • Understanding settings for model training
    • Data processing agreements and retention: your checkpoints
  2. Module 2

    Features with security impact

    • Projects, Artifacts and shared content
    • Connectors and MCP: access to email, calendars, drives
    • Web search, file uploads and code execution
  3. Module 3

    Assessing risks realistically

    • Prompt injection via web pages and documents
    • Hallucinations and source checking
    • Live demo: a manipulated file in context
  4. Module 4

    Data classes and processes

    • Traffic-light model for your information
    • Exercise: assess use cases from your daily work
    • Approval process for new connectors
  5. Module 5

    Usage policy

    • Essential content and wording
    • Roles of IT, data protection and business units
    • Joint draft

Why Claude needs its own rules

Claude is much more than a chat window. Through connectors it can access email, calendars, knowledge bases or ticketing systems, and with code execution it creates files and analyses. That greatly increases its value but also widens the attack surface.

A generic AI policy is therefore often not enough. You need clear rules on which integrations are allowed and who approves them.

GDPR and the EU AI Act

We show which documents and settings your data protection team should review and which questions arise with personal data. We also explain the AI literacy obligation under Article 4 of the EU AI Act, which applies to organisations deploying AI systems.

The workshop provides a basis for decisions but is not legal advice.

  • Which plan tier fits your requirements?
  • Which data may go in, and which never?
  • Who approves connectors and who monitors them?

Claude compared

Many companies use Claude alongside ChatGPT, Microsoft 365 Copilot or Google Gemini, so your rules should work across tools. We show where Claude differs and where the same principles apply, so you do not need a new rulebook for every tool.

Frequently asked questions

Does Anthropic use our data for training?

It depends on the plan and settings; commercial plans have different terms than consumer accounts. We show where to find the current terms and how to assess them for your company.

Are connectors inherently risky?

Not inherently, but every connector extends what Claude can access. With a clear approval process and least-privilege access, the risks are manageable.

Who benefits most from this workshop?

Companies that want to introduce Claude or have found that employees already use it privately. It also complements existing ChatGPT or Copilot policies.

Workshop

Get in touch

Want to introduce Claude safely with clear rules for your team?

Tell us a little about your team – we will get back to you within 24 hours with a proposal for content, format and dates.

Prefer to talk?

We only use your details to handle this request. See our privacy policy.