What guides us
Responsible AI
These are the rules we hold ourselves to in every system we design, for a client or for ourselves.
Our rules
- 01
Lawful, and more than lawful.
Every project is assessed on three questions: is it lawful, is it ethical, and does it truly benefit the people it affects. All three must be answered yes.
- 02
No deception.
Our systems do not pretend to be human, do not hide what they are doing with a person's data and do not use design tricks to push a decision.
- 03
Least data.
We collect what the task needs and no more, we keep it for a stated time, and we design deletion that works.
- 04
People decide.
Software may read, sort and recommend. Decisions about people stay with people, and those affected can ask for a human review.
- 05
Fairness is tested.
Where a system ranks or filters people, we test it for unfair differences between groups before it is used and while it is used.
- 06
Explainable in plain words.
If we cannot explain what a system does and why to the person it affects, we do not ship it.
- 07
Security by default.
Data is encrypted, access is limited to those who need it, and every access that matters is recorded.
- 08
We say no.
We do not build systems for surveillance of individuals, for manipulation, or for any purpose that conflicts with these rules.
In practice
Copper Hire, our own product, is built on these rules: employers see what a candidate can do before they see who the candidate is, and the candidate decides who learns their identity.
Visit copperhire.com (opens in a new tab)