Research and responsibility
Evidence-aware, not evidence-theatre
We separate what a system is designed to support from what has been independently demonstrated—and we keep that distinction visible.
Intended use versus evidence
Intended use is a design statement: what the tool is built to help someone do. Evidence is a different claim. It requires a method, a population, and a result that can be inspected.
USLC will describe intended use in plain language. We will not dress that description up as scientific validation, peer review, or guaranteed performance.
What we will not claim
We will not claim that a system identifies people, reads personality from appearance, diagnoses a condition, or decides who should be hired, promoted, or paid.
We will not publish “forty percent faster” style results, awards, or client logos without a source we can stand behind. If you see numbers like that on older pages or other sites, do not treat them as current USLC evidence.
How we review public copy
Product, privacy, and commercial language should agree with one another. A marketing sentence that contradicts the acceptable-use rules or the AI disclaimer does not ship.
Educational articles will show an author or owner, a date, and a distinction between evidence, interpretation, and opinion once those pieces exist.
Open questions
We still owe the public named policy owners, confirmed commercial terms, and a fuller account of which evaluation methods we use inside each system. Those gaps are listed in the Trust Center rather than filled with guesswork.
If you are a researcher or practitioner who wants to discuss methods, use the contact form. Bring the question, not a request for a fabricated endorsement.
Evidence questions
Are USLC systems scientifically validated?
We do not publish that claim. Intended use is described on each system page. Independent validation would be named, dated, and scoped if it exists.
Why avoid performance percentages?
Because unsupported numbers travel farther than caveats. We would rather under-claim than invent a result.
Where are AI limits written?
In Responsible AI, the AI Disclaimer, and Acceptable Use. System pages add one plain-language sentence and a link.
Can organisations request a methods conversation?
Yes. Use the sales or general contact route and describe the evaluation question you need answered.
Read the practice, not just the principle
Responsible AI explains how assistance is supposed to work. Our Systems shows what each product is for.