Responsible AI at Houser

Houser answers questions about housing regulations: what the rule says, where it comes from, and what has superseded it. It does not screen applicants, score tenants, rank households, or produce eligibility or termination determinations, and it is not designed to. Our Terms of Service state this plainly, and our Acceptable Use Policy prohibits using any output as the sole basis for an adverse action, eligibility determination, or other decision affecting an individual. Those decisions belong to qualified humans, reviewed and signed by the people your programs make responsible for them.

Each factual claim in a Houser answer cites its source: the CFR section, PIH or Housing Notice, IRS guidance, QAP, or state statute it came from. Citations open the document side by side with your conversation so you can read the actual language in context. When guidance has been superseded or rescinded, Houser flags it and points to the current authority instead of quietly serving the old one.

Before an answer reaches you, an automated verification pass checks its claims against the retrieved sources, and answers carry a confidence signal when review is especially warranted. And when our knowledge base does not contain the answer, Houser tells you that directly. An honest “I don’t know” beats a confident guess in this line of work.

Houser never auto-signs, auto-submits, or files anything. Drafting features produce drafts for your review, and correspondence features include explicit human send gates, so a person you authorize reviews and approves before anything leaves your organization. Our terms require that a qualified human review every output before it is used. The product is built to make that review easy, with the sources one click away.

Houser’s knowledge base is built from public U.S. federal and state housing materials: the CFR, PIH and Housing Notices, IRS guidance, published QAPs, state statutes, and similar sources. We track each document’s status, active, superseded, or rescinded, so currency is part of the answer, not an afterthought. For how we handle your data, including the identifier screening that runs before anything reaches an AI model and our full subprocessor list, see Security at Houser. We do not use your content to train AI models.

Responsible AI is a shared arrangement. We ask three things of every user: verify output against the cited authority before relying on it, never use output as the sole basis for a decision about a person, and remember that Houser provides general regulatory information, not legal advice. The full commitments are in our Acceptable Use Policy and Disclaimers.

We are a Colorado company, and Colorado has been first to regulate AI in consequential domains like housing. Senate Bill 26-189, effective January 1, 2027, governs automated decision-making technology that materially influences consequential decisions, and House Bill 26-1263 sets transparency standards for conversational AI. Houser’s design already reflects the principles behind both: it is offered and licensed strictly as decision support, our Acceptable Use Policy prohibits deploying it to make consequential decisions, and every AI surface in the product tells you that you are working with AI.

Many of our customers are housing agencies with obligations of their own under these laws. For Enterprise customers, we provide documentation during contracting that supports those obligations, including plain-language descriptions of what Houser does and does not do that you can use in your own notices. We are following the Colorado Attorney General’s rulemaking closely and will keep our practices, and this page, current as the rules are finalized.

If you have questions about any of this, or your agency needs documentation for a procurement or compliance review, email support@aihousers.com. We answer these ourselves.