Title VII of the Civil Rights Act of 1964 prohibits employment practices that discriminate on the basis of race, color, religion, sex, or national origin. The Equal Employment Opportunity Commission's technical assistance on algorithmic decision-making applies that existing law to software: a resume screener, a video-interview scorer, or a ranking model is an employment "selection procedure," and it is judged by its outcomes.
Two points matter most for a right to freedom from algorithmic discrimination. First, intent is not the test. Under the disparate-impact framework, a neutral-looking tool that selects one protected group at a substantially lower rate than another can be unlawful regardless of whether anyone set out to discriminate — which is why "no protected attribute is in the training data" is not a defence. The Uniform Guidelines on Employee Selection Procedures supply the familiar four-fifths rule of thumb for spotting such a gap, though the EEOC treats it as a rough screen rather than a safe harbour.
Second, buying the tool does not transfer the liability. The EEOC's position is that an employer is generally responsible for a vendor-supplied algorithm's adverse impact, including where the vendor administers it. That places a duty to test before deployment on the organization that will use the system, not only on the company that built it.