
No-training defaults, EU hosting, and self-hosted models - the three tiers of business AI privacy, what each costs, and which your data actually needs.
"Private AI" means three different things depending on the tier: consumer plans may train on your chats, business plans contractually do not, and true isolation (EU hosting or self-hosted models) is reserved for regulated data. Most small businesses need the middle tier - a contractual no-training guarantee - which costs nothing extra on the right plan. You only pay for real isolation when GDPR, client confidentiality or trade secrets make it mandatory.

For the next step after private ai for business, review AI Vendor Security Questionnaire: Ask for Evidence Before Access, and AI Data Processing Agreement Checklist for Business Buyers, and GDPR-Compliant AI Tools for Business: Check the Whole Use Case.
For the current product boundaries, visit the Krater private AI hub. Its linked guides cover private AI searches, consumer, API and team training policies and private AI chat controls without treating no-training as a retention guarantee.
Consumer chatbots have trained on user chats by default at various points - opt-outs exist but are buried and do not apply to everything. Business tiers changed this contractually: OpenAI's Business/Enterprise, Anthropic's Team/Enterprise and Krater all carry a no-training default, meaning prompts and files are excluded from model training as a matter of contract, not a settings toggle an employee can miss.
Private deployments can pair local and cloud providers through Open WebUI; the Open WebUI alternative page compares its self hosted setup with managed access.
No-training is half the story. Also check: how long chats are retained, where data is processed (US-only vs EU options), which subprocessors touch it, and whether admins can see or export team usage. Two vendors with identical "we don't train" language can differ on all of the above.
A team inventory of employee-paid tools is outlined in Shadow AI, with migration steps for one approved workspace.
Use this enterprise AI platform scorecard to evaluate data boundaries, administration, and model access.
Msty's desktop app chats with local models or connected API providers; the Msty alternative comparison details its local control against hosted access.
Turn these privacy principles into an AI policy template that employees can follow.
If you are evaluating Anthropic for a larger organization, see the details on Claude Enterprise pricing and separate usage billing.
Running Llama or Mistral on your own infrastructure is the only architecture where no vendor touches the data. It is also a real operations job - GPUs, updates, evals - that is a poor trade below roughly fifty daily users or without a hard residency requirement. For everyone else, a business tier with a DPA gets 95% of the protection for none of the ops.
Thirty minutes with Adam Johansson, an AI expert who maps out what actually fits your business: the right tools, plan and setup for your team size and usage. Free, no strings attached.
Free for any business, any size. We reply within one business day.

Adam Johansson
AI Advisor, Krater
Privacy in business AI is mostly a purchasing decision, not a technical one: put the company on a business tier with a contractual no-training default, write down what may never be pasted into any AI, and reserve isolated or self-hosted deployments for the data that genuinely needs them. That posture costs little and covers nearly everything an SMB handles.
Policies checked October 2026 on vendor terms and pricing pages.