The Endpoints You Published Without Knowing.
Exposed AI infrastructure is now an attack surface class.

Does your organisation know how many model endpoints it is running — or who stood them up? A passive external assessment is where the honest answers begin: what you expose to the outside world, and who owns that risk.

Assess Your Exposure →

Two years ago, an organisation’s AI exposure was a procurement question: which vendor, which contract, which data residency clause. In 2026 it is an attack surface question, because a meaningful proportion of the AI infrastructure inside Australian organisations was not procured at all. It was installed.

A data scientist stands up a local model server to avoid sending client material to a third party — a genuinely good instinct. A developer runs an inference gateway on a cloud instance to prototype a feature. A team wires an agent framework to an internal document store so it can answer questions about the knowledge base. Each of these is a reasonable act by a competent person. None of them typically involves a change request, and a surprising number end up answering on a public IP address with no authentication in front of them.

The core problemMost model-serving software is designed for a trusted local network and ships accordingly — bound to all interfaces, no authentication by default, an administrative API alongside the inference API. That is a sensible default for a laptop. It is a catastrophic default for a cloud instance with a public address.
The surface nobody is scanning
0
serving, gateway, MCP and RAG platforms now fingerprintable from outside
0
MCP servers found exposed with zero authentication
0
to stand up a model server; longer to convene a review board
0
signatures conventional vulnerability scanners have for any of it
Sources: open-source AI security tooling surveys (2026); Trend Micro MCP exposure research; Praetorian LLM infrastructure fingerprinting.

What is actually appearing on the perimeter

Security tooling released through 2026 now fingerprints more than sixty distinct model-serving, gateway, MCP and RAG platforms on public endpoints. The categories worth understanding:

Local model servers

Software that runs open-weight models on your own hardware. The value proposition — data never leaves your environment — is real, and is exactly why privacy-conscious teams adopt it. The failure mode is that the environment is not as closed as assumed. An exposed server typically permits anyone to enumerate loaded models, submit arbitrary prompts, and in many configurations pull or delete models entirely.

Inference gateways and proxies

Middleware that routes requests to multiple model providers, handles rate limiting and centralises billing. Because it centralises, it also holds provider API keys. An exposed gateway is a credential store with an HTTP interface, and the resulting bill is a secondary concern next to the fact that a third party is now making requests attributed to your organisation.

MCP servers

The Model Context Protocol became the connective tissue of agentic systems during 2026. An MCP server exposes tools to a model — file access, database queries, ticketing, code repositories, internal APIs. The security property is the important one: an MCP server is, by design, a remote execution surface. Its purpose is to let something else take actions in your environment. If it is reachable without authentication, you have published an action interface to your internal systems.

RAG backends and vector databases

Retrieval-augmented generation requires an index of your documents. That index is a derivative of the source corpus — and an exposed vector store frequently permits similarity queries that reconstruct meaningful portions of the underlying content, without any access control ever being tested. Organisations that carefully classify a document repository routinely give no thought at all to the embedding index built from it.

Four platform classes — what each one gives away
Local model servers
Ollama, vLLM, LM Studio and similar
Allinterfaces bound by default
Noneauthentication by default
AdminAPI alongside inference API

Why it gets deployed: data never leaves your environment. A genuinely good instinct, and exactly why privacy-conscious teams adopt it.

What an attacker gets: enumerate every loaded model, submit arbitrary prompts at your compute expense, and in many configurations pull or delete models entirely.

Why it is missed: the environment is not as closed as assumed. The failure is not the software — it is the assumption that the instance is private.

Gateways and proxies
Routing, rate limiting, centralised billing
Holdsprovider API keys
Centralby design
Billedto you

Why it gets deployed: it centralises routing across multiple model providers and handles rate limiting and cost attribution.

What an attacker gets: because it centralises, it holds provider API keys. An exposed gateway is a credential store with an HTTP interface.

The real damage: the bill is the secondary concern. The primary one is that a third party is now making model requests attributed to your organisation.

Model Context Protocol servers
The connective tissue of agentic systems
492found exposed with no auth
9.1CVSS on one major package
Remoteexecution by design

Why it gets deployed: it exposes tools to a model — file access, database queries, ticketing, code repositories, internal APIs.

The security property that matters: an MCP server is, by design, a remote execution surface. Its entire purpose is to let something else take actions in your environment.

What an attacker gets: if it is reachable without authentication, you have published an action interface to your internal systems. Not a data leak — a lateral movement path that an autonomous agent can use exactly as intended.

Retrieval backends and embedding indexes
The derivative nobody classified
Derivedfrom your document corpus
Rarelyclassified
Similarityqueries reconstruct content

Why it gets deployed: retrieval-augmented generation requires an index of your documents.

What an attacker gets: an exposed vector store frequently permits similarity queries that reconstruct meaningful portions of the underlying content, without any access control ever being tested.

The governance failure: organisations that carefully classify a document repository routinely give no thought at all to the embedding index built from it. Whatever classification applies to the source applies to the derivative.

Select a class to see why it is deployed and what it exposes

Why this surface is being missed

Three structural reasons, and none of them is incompetence.

It falls between two functions. Security teams scan for known application and infrastructure signatures. AI governance functions review model use, bias, explainability and vendor terms. An unauthenticated inference endpoint is neither a traditional application vulnerability nor a model governance question, so it sits in the gap between the two workstreams.

The deployment is fast and the review cycle is not. A model server can be running in ten minutes. Architecture review boards do not convene in ten minutes, and the person who deployed it did not think of it as infrastructure — they thought of it as a tool.

The scanners do not know what they are looking at. Conventional vulnerability tooling has no signature for these platforms. An exposed inference API frequently returns a clean result, because there is nothing to detect against — it is not a vulnerability, it is a working service doing exactly what it was configured to do.

The governance consequenceAn organisation can hold a defensible AI governance framework — an approved-use register, vendor assessments, an ISO/IEC 42001 alignment programme — and still have three unauthenticated model endpoints on its public perimeter that appear in none of it. The framework governs what was declared. Nothing governs what was not.

How it maps to obligations you already carry

No new regulation is required to make this a compliance problem. The existing stack covers it:

  • CPS 234 — an information asset is an information asset regardless of how novel it is. An undeclared model server holding or indexing regulated data is unclassified and unprotected by definition.
  • Essential Eight and the ISM — exposure minimisation and administrative interface restriction apply directly. An open management API on a model server is an unrestricted administrative interface.
  • Privacy Act, APP 11 — if personal information reached an unauthenticated endpoint, reasonable steps were not taken. If it reached an embedding index nobody classified, the same conclusion follows.
  • ISO/IEC 42001 — an AI management system requires a defined scope. Undiscovered systems are outside scope by accident, which is not a defensible basis for a scoping decision.

The agentic dimension

2026 was the year agentic AI moved from prediction to operational concern — a readership poll of security professionals put agentic and autonomous systems ahead of deepfakes and ransomware as the top anticipated attack vector for the year. The defensive analysis that has held up best makes a specific point about tempo: reconnaissance is now continuous and automatic rather than a discrete pre-attack phase, and a blocked attempt resumes as soon as the agent adapts.

That has a direct implication for exposed AI infrastructure. An endpoint that appears on a Tuesday afternoon during a prototype does not need to survive until the next quarterly scan to be found. Continuous automated discovery means the window between exposure and enumeration is now measured in hours.

It also means the endpoint is not merely a data risk. An exposed MCP server connected to internal tooling is a lateral movement path that an autonomous agent can use as designed — no exploit, no malware, no boundary crossed. The same pattern as June’s enterprise application campaigns, applied to a newer class of system.

How much shadow AI do you have?
Tick every statement that is true, or that you cannot rule out.
Someone in our organisation has run a local model server on a cloud instance in the last twelve months.
We use an inference gateway or proxy that holds provider API keys.
We operate at least one MCP server, and we could list the tools it exposes.
We have a vector store or embedding index built from internal documents.
Any of the above was stood up without a change request.
We have AI features enabled by default in SaaS we already bought, and we have not inventoried them.
Our AI governance register was assembled from what teams declared, not from an external scan.
We could not say, today, how many model endpoints we run.
Tick every statement that is true of your organisation

What to do, in order

  • Discover before you govern. Run a passive external enumeration specifically looking for model-serving, gateway, MCP and vector-store signatures across your perimeter and your cloud address space. You cannot write policy for systems you have not found.
  • Ask the question internally, without consequence. A short, explicitly amnesty-framed request to engineering and data teams — “tell us what you have running, nobody is in trouble” — surfaces more in a week than a control review does in a quarter.
  • Treat MCP servers as privileged infrastructure. Authentication, network restriction, and an explicit inventory of the tools each server exposes. Anything less publishes an action interface.
  • Classify the index, not just the source. Whatever classification applies to a document repository applies to any embedding derived from it.
  • Add AI infrastructure to the asset register as a distinct class. Not under “software.” It has a different failure mode and a different owner.

This is a surface almost nobody in the Australian mid-market is currently assessing. That is precisely why it is worth assessing now, while the finding is still cheap to remediate rather than expensive to disclose.

Passive only — no systems accessedBlackFlag Advisory identifies exposed AI infrastructure using publicly available data and passive observation only. No endpoint is queried, prompted, enumerated or tested. Findings are mapped to CPS 234, the Essential Eight, the ISM, the Privacy Act and ISO/IEC 42001.
SourcesOpen-source AI security tooling surveys and public repositories cataloguing model-serving, gateway, MCP and RAG platform fingerprinting (2026); Dark Reading readership research on anticipated 2026 attack vectors; Barracuda Networks analysis of agentic attack tempo; Palo Alto Networks 2026 forecasts; APRA CPS 234; ASD Essential Eight and Information Security Manual; ISO/IEC 42001. Analysis by BlackFlag Advisory.

Shadow AI Does Not Appear
In Your Asset Register.

A BlackFlag Advisory passive assessment identifies model servers, inference gateways and agent endpoints exposed on your public perimeter, and maps them to your AI governance and information security obligations.

Request an Assessment →
Passive Only — No Systems Accessed

All BlackFlag Advisory assessments use exclusively passive OSINT techniques and publicly available data sources. No systems, networks, or accounts are accessed, probed, or tested at any time. Board-ready output delivered within three to seven business days.