Security

How CanvasKB keeps your company knowledge secure: access-aware answers, no public model training, and insight that shows patterns, not people.

Security & Trust

Your documents. Your permissions. Your AI.

Trust shouldn’t be something you take on faith. Here’s exactly how your company’s knowledge stays private, who can see what, and the lines we never cross — in plain terms, not fine print.

You control who sees what
Never used to train a public AI model
Reports show trends, not people
Your data stays under your control

How it is protected

Three guarantees, built in

Access control
You control who sees what

CanvasKB only answers using documents a person could already open on their own. We don’t grant new access, and we don’t create shortcuts around the permissions you’ve already set up.

  • Works within your existing permissions — nothing new to configure
  • Never opens a door that wasn't already open
Private AI
Your AI, your environment

Every answer is generated on private enterprise AI, not a shared public AI model. Your documents are never used to improve a model that other companies use.

  • Private enterprise AI, not a shared public model
  • Never trains any AI model outside your company
Anonymized insight
Insight for leaders, privacy for employees

Leaders get visibility into where knowledge gaps are slowing teams down — by topic and team, never by name. No employee is ever scored, ranked, or watched.

  • Shows patterns and trends, not individuals
  • No individual tracking, ever

Security FAQ

Straight answers for your security review

Plain answers to the questions security teams, IT, and business stakeholders ask us most — before a contract, not after.

Access
Permission-aware answers
Question:
Could someone see a document they don't already have access to?
Answer:
  • No. CanvasKB only answers using documents a person could already open themselves. It never creates a workaround for existing permissions.
Privacy
No public training
Question:
Will our documents be used to train your AI for other customers?
Answer:
  • No. Your content is never used to train any public or shared AI model — not for us, and not for anyone else.
Oversight
Anonymized insight
Question:
Can a manager see what an individual employee asked?
Answer:
  • No. Managers only ever see trends and topics across a team — never who asked what, or how often.

Compliance

Built for buyers who answer to auditors

We know your security and compliance team will have questions before anyone signs a contract. Here’s exactly where we stand today — no overstating, no fine print.

SOC 2

Type II

independent audit in progress

SOC 2 is the security standard most enterprise buyers use to check how a vendor actually handles data — not just what they claim. Our independent audit is currently underway, and we’ll share results as soon as it’s complete.

Private

Deployment

available now

For regulated industries and EU-based teams that need their data to stay inside a dedicated environment, private deployment is available today.

“If a person doesn’t have access to a document, CanvasKB will not use it to answer them.”

CanvasKB security principle

  • Permission-aware by default

We only answer from what you can already see.

  • Never used to train a public AI model

Your documents never train someone else’s AI.

  • Privacy for your people

What your team asks stays private — even from their own manager.

  • No individual scoring

No one is ever scored, ranked, or watched.

  • Your data stays in your environment

Your documents are never copied out of your control.

  • No invented answers

We don’t guess. If the answer isn’t in your documents, we won’t invent one.