ACLU Civil Rights Complaint Against Criteria Corporation (California Civil Rights Department)

This is an administrative complaint, not a lawsuit. The complainants ask CRD to investigate how Criteria develops, sells and administers its hiring assessments, and to initiate a director’s complaint. The central focus is Emotify, which Criteria markets as measuring a candidate’s ability to perceive, understand and manage emotions. The complaint compares Emotify’s tasks to clinical tools used to assess autism, including NEPSY-II, SRS-2 and MIGDAS-2.

Who is the plaintiff?

The complainants are the ACLU Foundation and the ACLU Foundation of Northern California. Attorneys from both organizations signed the complaint, which is addressed to CRD Director Kevin Kish.

What is being alleged?

The complaint pleads three overlapping theories under FEHA:

1. Emotify is a prohibited pre-offer disability-related inquiry or medical or psychological examination.
2. Emotify tends to screen out autistic applicants without a job-related justification. The complaint points to Emotify’s three-second time limits, its low-intensity facial expressions and its scoring against a general-population norming group.
3. Criteria and the employers using Emotify are likely not meeting their reasonable accommodation obligations.

The complaint also raises disability, race and sex concerns about other Criteria products: Cognify, Interview Intelligence, AI Scoring, AI Proctoring and its personality assessments.

Why is the defendant being named?

The complaint argues that Criteria acts as an agent of employers under FEHA because it administers assessments that evaluate and screen applicants and make recommendations about them. It relies on Raines v. U.S. Healthworks Medical Group and Mobley v. Workday to support this agent-liability theory.

For Employers and HR Teams

The complaint highlights that pre-offer assessments may face scrutiny if they resemble clinical or psychological tests, even when they are used outside clinical settings. Organizations using emotional intelligence, personality, game-based or AI-scored assessments may need to consider three questions. Does each tool measure competencies tied to the specific role? How are scores normed? How can applicants request accommodations or alternative assessments before they are screened out? The case also increases the importance of reviewing vendor accessibility materials and documenting job-relatedness.

For HR Technology Vendors

The complaint tests whether an assessment provider can face direct FEHA exposure as an agent of employers, drawing on Raines and Mobley v. Workday. It highlights how product design choices may be scrutinized as evidence of disparate impact, including time limits, stimulus intensity, norming groups and rigid response formats. The complainants also cite Criteria’s marketing claims, employer-facing accessibility guidance, candidate accommodation workflows and developer fairness research. Providers may need to consider whether their candidate information, accommodation mechanisms and validation evidence are ready for regulatory review.

Provider

Criteria Corporation

Products or Systems Named

• Emotify: an emotional intelligence assessment and the primary focus of the complaint
• Cognify: a game-based cognitive aptitude assessment
• Interview Intelligence (IIQ): video interview features, including AI-generated summaries and AI-assisted question creation
• AI Scoring: an add-on that automatically scores interview transcripts
• AI Proctoring: used with Criteria’s pre-employment assessments
• Employee Personality Profile (EPP)
• Criteria Personality Inventory (CPI)

Employment Use Cases

• Pre-offer, pre-employment assessment of job applicants
• Emotify, according to Criteria’s marketing as quoted in the complaint:
• • Marketed as suitable for any role
  • Recommended especially for managers, future leaders, graduate recruitment, sales, customer service and roles involving emotional labor
• Emotify, according to the complainants:
• • Also used for roles such as software engineer, financial analyst and website designer
• Emotify industries cited from Criteria case studies: financial services, healthcare, insurance, human resources, professional services, digital marketing and automotive
• Video interviewing with AI-assisted summaries, question creation and transcript-based scoring
• Automated proctoring of pre-employment assessments

Role in Decision-Making

The complaint alleges that Criteria markets and administers products through which it evaluates and screens applicants and ultimately makes recommendations about them. Emotify scores are normed against a general-population group and reported as percentile rankings from 0 to 100, classified relative to an “average” level. The complaint alleges that, where Emotify is used as an initial screening device, applicants may be rejected before they can request an accommodation or an alternative assessment.

Alleged Harm

The complaint makes the following allegations:

• Pre-offer inquiry: Emotify elicits disability-related information before a job offer.
• Screening out: Emotify disproportionately screens out applicants with autism and related neurodevelopmental conditions for non-job-related reasons.
• Accommodation: Applicants receive too little information to know whether to request an accommodation.
• Employer-facing guidance: Criteria’s accessibility guide tells employers that Emotify may not be suitable for autistic applicants or applicants with Developmental Coordination Disorder, ADHD or Tourette syndrome, but Criteria does not give applicants this information.
• Cognify: An early version of Cognify showed race and gender disparities in a 2022 study by its developers.
• Speech and video AI: AI speech and video technologies may perform worse for disabled applicants, people with accents, and Black, Latine and Indigenous speakers.
• Personality assessments: These may disadvantage people with autism, depression or anxiety.

Data or Proxy Variables Discussed

• Images of facial expressions shown at varying intensities
• Response time limits (3 seconds per round in “Matching Faces”; 1 minute per round in other tasks)
• Rule that the same image cannot be used for more than one scenario in “Emotional Ties”
• Norming against a general-population (“global”) group, with percentile rankings
• Interview transcripts used for automated scoring
• Image capture, image analysis, and screen and audio sampling in AI Proctoring
• “Big Five” personality traits, including “Stress Tolerance” / Emotional Stability