Most acceptable use policies tell users what they may not do with a vendor’s system. This one runs the other way: it is what LFW may and may not do with yours.
Version 1.0, October 6, 2026. Part of the published practices on the trust page. The services agreement is the contract these rules sit under.
1. Client data
LFW uses a client’s data only to do the work the client engaged it for. It does not sell it, use it to build profiles, or use it for a purpose the client has not asked for.
Confidential information is kept with at least reasonable care, shared only with people who need it and are bound to the same terms, and returned or destroyed within 15 days on request or at the end of the engagement, except for copies in encrypted backups that expire on schedule.
LFW keeps client data in the client’s systems and avoids holding copies on its own equipment. Work in progress lives in the repository the client owns.
2. Client systems
LFW signs in to a client’s systems only through accounts the client granted, uses them only for the work, and changes nothing outside the agreed scope without asking first.
LFW does not read, change or download data on a client’s system that the work does not need. If it reaches such data by accident, it stops and tells the client.
Information a client’s own system serves to the public without credentials is not treated as confidential, but LFW handles apparently sensitive exposed material with discretion, reports the exposure promptly so it can be closed, and does not use or share its contents.
3. Credentials
LFW never asks a client to send a password in email or chat, and never shares an account. Each person who needs access gets their own account; a shared credential, where a platform forces one, goes in the password manager vault for that client and nowhere else.
LFW turns on multifactor authentication on every delegated account where the platform offers it, and does not reuse a credential across clients or systems.
4. AI tools
We use AI tools, including Claude, to draft and check work. Client data is never used to train models, and a person reviews every output before it reaches you.
AI tools are assistants to LFW, not a party to the work: LFW stays responsible for every output. They are never given a client’s credentials, and they are not used with any data a client has named in its contract as excluded.
The AI providers LFW uses are listed as subprocessors on the trust page, with what each one sees. A client can ask LFW not to use AI tools on its work, and LFW will say in writing what that changes.
5. LFW’s own equipment and accounts
Working credentials live in a password manager, never in plain files or in the repository. Builds that run third-party code run in a disposable container with no keys and no private files, so a poisoned dependency cannot read them.
Private client files and LFW’s own records are kept outside the code repository and never enter a build. LFW’s server access is by key, from LFW’s own equipment only.
6. Automated tools and scanning
LFW’s own scanners read only what a public visitor could, on sites LFW is engaged on or that are public, within request limits, and never attempt to reach private systems. A client’s site is scanned under the engagement that covers it.
LFW does not run denial-of-service tests, credential guessing or social engineering against anyone’s systems, including its own clients’, unless the client has asked for a test in writing and set its terms.
7. Subcontractors and other people
No subcontractor works on a client’s site without the client’s written approval. Anyone LFW brings in is bound to this policy and to confidentiality terms at least as protective as the services agreement’s.
8. Breaches of this policy
If LFW breaks one of these rules, that is an incident under the incident response plan: the client is told, the record is written, and the practice changes.
Questions about this policy: hello@lfw.com.