HumaniCore AI Debuts Independent Audits and Certification Standard for AI-Enabled People Decisions

Pre-revenue on purpose, the company built its methodology on evidence before selling it: audits that count every missing record and state whether it could change a fairness result.

LAS VEGAS, Oct. 20, 2026 (GLOBE NEWSWIRE) -- HumaniCore AI today introduced independent third-party audits for AI-enabled people decisions, and the certification standard behind them, for both the vendors that build these tools and the enterprises that deploy them. The company is exhibiting at the HR Technology Conference in booth SU93 and will be pitching in the conference’s Pitchfest on Wednesday, October 21, at 5:00 PM.

Employers remain responsible for employment decisions made with AI, regardless of who built the tool. Researchers who reviewed all 116 bias audits published under New York City's Local Law 144 found that 83 percent reported missing race or sex data for some applicants. Under a worst-case assumption about those missing records, 70 percent of reported results could fall below the four-fifths benchmark long used to flag possible adverse impact (one group selected at less than 80 percent of the rate of the most-selected group); under a moderate assumption, 28 percent. In one compliance review of 391 employers covered by the law, only 18 had posted an audit.

Pre-revenue on purpose and rather than rush a seal to market, HumaniCore AI took the time, methodically and internally, to make sure every audit produces evidence that is defensible when examined. HumaniCore AI tested its methods on public census data: more than 250,000 records across hundreds of audit runs, and 240 controlled trials in which the range it reports contained the true value every time. Five of its methods are patent pending as of October 8, 2026. Along the way, the company declined revenue twice when accepting it would have compromised its independence.

HumaniCore AI certifies on two tracks. Vendor certification covers the AI-enabled solutions that product companies build, such as applicant tracking systems, performance management systems, compensation tools, skills assessors, and others. Enterprise certification covers how an organization uses its full HR tool stack wherever AI shapes decisions about people, across hiring, promotion, evaluation and termination. The engine is deterministic: the same records, run against the same rule-pack version, produce the same verdict every time, so any third party holding the same inputs can reproduce it. Where demographic data is incomplete, the report classifies why each value is missing and reports a range rather than a single number the data cannot support. Each certification also names the party entitled to rely on it, a requirement absent from the bias audit rules in force in the United States today.

“AI in HR should make people decisions more human, not less, and today very little of it is independently checked,” said Haley Friary-Schenk, founder and chief executive officer of HumaniCore AI. “I spent twenty years leading global HR and implemented robust HR technology stacks five times. When AI entered the stack, I saw that HR now had to defend decisions nobody could explain.”

“We are pre-revenue on purpose,” Friary-Schenk said. “We built HumaniCore AI to provide oversight for every tool behind AI-enabled people decisions with evidence and discipline, and we were not going to put our seal on anyone’s website until we could defend every verdict behind it. Accountability is what makes speed safe. Certification is how you get both.”

Each engagement is designed to produce an audit report stating what was found and what could not be determined, a certification determination with a verifiable seal, and an evidence file, followed by annual recertification that measures what moved between audits. Every seal will carry a verification reference, so the party entitled to rely on it can confirm current status. HumaniCore AI never fixes what it checks, and certification decisions sit outside vendor influence by governance design.

The launch follows another recent change in state law. Connecticut Public Act 26-15, which took effect October 1, 2026, provides that an employer’s use of automated employment decision technology is not a defense to a discrimination complaint, permitting a court or the state commission to consider documented anti-bias testing. New York City, California, Texas and Illinois requirements are already in force, and European Union high-risk requirements for AI used in employment apply from December 2, 2027.

HumaniCore AI has delivered schema assessments to two vendors. The company was a top-five-percent semifinalist among 531 teams in the Texas A&M AI Venture Velocity Challenge, received funding from Penn State’s Farrell Center to sponsor a six-engineer Penn State Learning Factory team and was selected for Invent Penn State, Venture Connection.

Product demonstrations run throughout the conference at booth SU93. Analyst and media briefings are available by appointment.

About HumaniCore AI

HumaniCore AI is the independent certification body for AI-enabled people decisions. The company is built to audit the AI tools organizations use to hire, promote, evaluate and terminate, starting with hiring, and certifies both the vendors that build them and the enterprises that deploy them. HumaniCore never fixes what it certifies, and its verdicts are reproducible by design. Patents are pending on five of its methods. Learn more at humanicoreai.com.

Note to editors: Trademarks and registered trademarks referenced herein remain the property of their respective owners.

MEDIA CONTACT

Jeanne Achille, The Devon Group for HumaniCore AI, jeanne@devonpr.com