Skip to content
All articles
Private AI

"Your data stays inside." Draw the arrows before you believe it.

"Your data stays inside" does not say which data, which part of the system, or inside what. A diagram of one request, with a mark on every arrow that crosses the perimeter, turns the sentence into a claim that can be tested.

BastionUpdated 3 min read

The ten-minute test: draw one request end to end

Draw the complete path of a single request: user, application, AI service, model, logs, storage, backup, update channel, support access. Do not skip a stage because it seems obvious. Then mark every arrow that leaves the building. The count is the answer.

Six arrows that can cross the perimeter without a decision

  • Telemetry and crash reporting are switched on by a default in a component nobody chose.
  • The automatic update channel is an outbound connection with write access to the system.
  • Backups frequently land somewhere with a different security posture than the primary.
  • A remote support tool is dormant until the day somebody needs it, and then it is fully privileged.
  • A retrieval or embedding service is hosted even though the model is local.
  • Logs contain prompts, and prompts contain the document.

A deployment can satisfy "the model runs locally" while five of those six arrows point outwards.

What a documented cloud service looks like, for comparison

A cloud provider that documents its flows properly makes this exercise easier. Microsoft's documentation for models sold by Azure states plainly that “The models are stateless: no prompts or completions are stored in the model”. It states separately that prompts and completions may be stored for abuse monitoring, where “Human reviewers assessing potential abuse can access prompts and completions data only when that data has already been flagged by the abuse monitoring system, or when the prompts and completions are part of a potentially abusive pattern of use”.

The first statement says what is not kept. The second describes two arrows, both documented and both conditional: storage for abuse monitoring, and a human reviewer's access to data that was flagged or that belongs to a potentially abusive pattern of use. A local deployment should be held to the same standard: ask the supplier to draw the diagram.

What staying inside does not protect

Even with every arrow pointing inwards, information can still leave through the front door. OWASP's 2025 Top 10 for LLM applications lists sensitive information disclosure as LLM02 and prompt injection as LLM01, and adds vector and embedding weaknesses as LLM08. Each of those is a route from an internal document to the wrong reader, wherever the model is hosted. Excessive user permissions, a compromised laptop, a malicious PDF and an unprotected log all survive the move indoors.

The version of the claim worth making

The useful formulation is narrower than the slogan, and it can be tested: inference traffic and model processing remain inside the organisation's controlled network boundary, with no outbound connection required to answer a request. It states what is controlled and implies what is not, and a security officer can verify it.

With Bastion

What Bastion changes

Bastion is a private AI system delivered as one sealed box that runs inside your building. One monthly fee covers the hardware, the model, the hardened operating system and support, and nothing your team types leaves the building.

Bastion answers with no outbound internet connection required, so the inference arrow does not cross the perimeter. Updates are signed and installed deliberately rather than pulled automatically, which keeps the update channel a decision rather than a default.

The arrows that remain are the customer's to decide: who may use the system, which documents it may read, where the logs go and who holds the backups.

Questions this article answers

What does "your data stays inside" actually mean?
Only what the architecture says it means. The testable version is narrower: inference traffic and model processing remain inside the organisation's controlled network boundary, with no outbound connection required to answer a request.
Which parts of a local AI system still send data out?
Six can: telemetry and crash reporting, the automatic update channel, backups, a remote support tool, a hosted retrieval or embedding service, and logs, which contain prompts, and prompts contain the document.
Does local processing prevent data leaks?
No. OWASP's 2025 list puts sensitive information disclosure at LLM02 and vector and embedding weaknesses at LLM08. Excessive permissions, a compromised laptop, a malicious document and an unprotected log all survive the move indoors.

On the record

  1. 1
  2. 2