Local does not automatically mean safe. A workstation running an OpenAI-compatible port can leak data if anyone on the network can call it, or if prompts are logged carelessly.
Network
Bind to localhost by default. Use VPN + reverse proxy when teammates need access. Avoid casual ngrok-style exposure.
Auth and audit
Require tokens even internally. Log who called which model alias, not full secret contents if policy forbids storing prompts.
Model and plugin supply chain
Download weights from agreed sources. Be careful with third-party Modelfiles that fetch remote content. Pin versions.
Data
Separate disks or folders for regulated corpora. Encrypt laptops. Define retention for chat logs.
Further reading on process: methodology, about, product context microai-one.
Write a one-page baseline and attach it to the install runbook. Security that exists only in someone's head disappears during staff changes.
Extra practice notes
For "A Security Baseline for Hosting Local LLMs on Office Workstations", triage feedback into slow, wrong, and confusing. Hold a short weekly pass. See MicroAI One and about. Size models with the VRAM table.
Endpoint inventory
Maintain a list of every OpenAI-compatible port on the machine. Shadow servers started for experiments are how accidental exposure happens.
Plugin risk
Browser extensions that talk to "local LLM" endpoints should be vetted. Malicious extensions can exfiltrate prompts.
Incident drill
Once a quarter, practice: revoke a token, rotate it, confirm old clients fail. Five minutes of drill beats a confused afternoon.
Physical access
Disk encryption matters when laptops leave the building with corporate PDFs embedded in indexes. Local AI expands the data footprint of a laptop.
More executable detail
For "A Security Baseline for Hosting Local LLMs on Office Workstations", create a single feedback channel where people report slow, wrong, or confusing behavior. Spend fifteen minutes a week tagging issues as hardware, prompt, permission, or expectation. Action after tagging beats endless model swapping. For orientation links you can cite MicroAI One and the about page.
Write decisions in a short runbook with dates. When something breaks, the first question is what changed; a dated note answers faster than memory. Include links to the exact model files and the ports you exposed. Keep the runbook next to the compose file or service unit so it cannot drift into a personal folder. (lp5-0)
Prefer boring defaults for anything shared. Power users can still override locally, but the shared alias should optimize for predictability. Predictability is what makes local AI feel like infrastructure instead of a hobby that resets every Monday. (lp5-1)
When you evaluate changes, change one variable at a time. If you swap the model, the quant, and the client on the same day, you will not know which change helped. Small experiments compound into a trustworthy stack. (lp5-2)
Published by MicroAI One (本地AI 1号网), a Shenzhen team working on local LLM hardware, software and deployment.