DFdom@fradley
← ~/writing
$ cat llm-endpoints-new-public-storage.md

LLM endpoints are the new public storage accounts

Last decade's example was the storage account: convenient, everywhere, reachable by default, and the subject of a breach headline every month for years. I wrote about hunting yours in July. The LLM endpoint repeats the pattern, with one difference for the attacker: they no longer need to find your data to profit from your misconfiguration. The endpoint itself can be resold, and it bills to you.

The trade has a name, LLMjacking, coined by Sysdig's threat research team. Their recent numbers: credential theft aimed at AI services rose 376% between the last quarter of 2025 and the first of 2026. Brokers validate stolen keys in bulk and resell them through reverse proxies on underground forums, and Sysdig put the potential burn on one hijacked frontier-model account at over $100,000 a day. GitGuardian counted 1.27 million AI-service secrets in public GitHub commits during 2025, up 81% on the year.

Azure estates are in the frame. Microsoft has been in court with Storm-2139, a group it says harvested exposed Azure OpenAI keys, wrapped them in a purpose-built web app, and sold access with the content filters stripped off. The abuse ran inside the victims' own subscriptions, on their invoices.

An inference endpoint is easier to monetise than a stolen virtual machine, which still has to be turned into money through cryptomining or resale as a proxy. It is metered compute that can be resold directly. It is also quiet: the traffic is API calls arriving at a legitimate endpoint inside a legitimate subscription, so nothing is exfiltrated and nothing falls over. On most estates the first detection control that fires is the invoice.

Three ways to host a model in Azure

The choice depends on whose model it is and whose GPUs run it.

  • Serverless, from the catalogue. Microsoft Foundry (Azure AI Foundry until June's rename) fronts the hosted-model market: the Azure OpenAI models, Anthropic's Claude family (generally available on Microsoft-hosted infrastructure since Build), Mistral, Llama and a long tail of others. You pay per token and Microsoft runs the GPUs. This is the right default for almost everyone, and the option this article secures: what you get back is an endpoint and a pair of API keys.
  • Managed compute. For open-weight models, Azure runs the model you choose on dedicated GPU instances you size, billed per hour rather than per token. You take on quota and capacity planning; you get model choice and a deployment that sits properly inside your network design.
  • Self-hosted on AKS. vLLM or similar on your own GPU node pools. It gives full control and the full operational burden, so it needs a specific reason: sovereignty rules the managed options can't meet, or token economics at a scale where dedicated hardware wins.

The demo path is the same for all three: a public endpoint and a credential in an app setting, stood up in an afternoon. The production controls have the same shape too, and none of them is AI-specific.

Demo-day defaults

An Azure OpenAI or AI Services resource arrives with key authentication on: two keys, no expiry, visible to anyone whose role can list them. Public network access is on by default too. Those are the two defaults that made the storage-account decade what it was, and the same two settings close them.

Kill the keys first. The resource takes disableLocalAuth, after which the only way in is an Entra ID token and a leaked connection string is just a string. Applications authenticate with their managed identity and a data-plane role (for inference, Cognitive Services OpenAI User), the same identity-is-the-perimeter move every other Azure PaaS service has been converging on. A scraper can't resell a managed identity.

Then take the endpoint off the internet: public network access off and a private endpoint in its place, with the DNS zone wired up so resolution stays inside your network. The account half of that in Terraform:

resource "azurerm_cognitive_account" "llm" {
  name                  = "oai-myapp-prod"
  location              = azurerm_resource_group.ai.location
  resource_group_name   = azurerm_resource_group.ai.name
  kind                  = "OpenAI"
  sku_name              = "S0"
  custom_subdomain_name = "oai-myapp-prod"

  local_auth_enabled            = false
  public_network_access_enabled = false

  identity {
    type = "SystemAssigned"
  }
}

The working example, with the private endpoint, the DNS zone, a model deployment and the role assignment for a caller identity, is in the blog-examples repo.

Two settings on one resource is hygiene. Azure Policy makes it governance: built-in policies exist for key access disabled, public network access disabled and private link required, so assigning them in audit mode makes the AI resources you didn't know about report themselves. The hackathon deployment from March is the one this catches.

Tripwires, because keys leak anyway

Some of your estate can't turn keys off tomorrow. Vendor tools still assume them, and a proof of concept always moves faster than its retrofit. For that tail, make the abuse loud.

A cost alert on every AI resource is the cheapest control here. LLMjacking is quiet in the logs and deafening on the meter, so a budget threshold that pages someone turns invoice-time detection into same-day detection. Send diagnostic logs to Log Analytics so you can see token volume by caller, and if you run Defender for Cloud, its AI services plan watches for this abuse pattern against the endpoints themselves.

This week

  1. Inventory the estate. One Resource Graph query lists every AI account and the two settings that matter:

kusto resources | where type == "microsoft.cognitiveservices/accounts" | project name, kind, resourceGroup, localAuthDisabled = properties.disableLocalAuth, publicAccess = properties.publicNetworkAccess

Anything with localAuthDisabled empty or false and publicAccess enabled is carrying demo-day defaults in production. 2. Disable local auth wherever nothing breaks, and rotate the keys on whatever remains. What's left becomes a named list with an owner. 3. Assign the built-in policies in audit mode. The report is your backlog, and the next hackathon lands inside the guardrail. 4. Put a cost alert on every AI resource, scoped tight and set low. Nobody notices a stolen key costing £40 a day inside a £2m bill; a per-resource budget does.

More useful tidbits coming, one a week.

discuss on linkedin → more writing