TRAIGA's Intent Standard: What It Actually Requires, and What It Leaves Exposed

Summary: Under Texas Business and Commerce Code section 552.056(c), disparate impact alone is explicitly insufficient to establish a TRAIGA violation for AI hiring tools. Liability requires intent to discriminate against a protected class. This is a genuinely employer-friendly feature of the statute, but it is narrower protection than it first appears, because it only insulates against TRAIGA itself, not against the federal discrimination statutes that operate on outcomes regardless of intent.

What section 552.056(c) actually says

TRAIGA prohibits developing or deploying an AI system with the intent to discriminate against a protected class under federal or state law. Section 552.056(c) makes the disparate impact point explicit: an AI hiring tool that produces skewed outcomes across a protected group, without having been designed with discriminatory intent, does not by itself establish a TRAIGA violation.

This is a deliberate policy choice, distinct from most other US AI hiring frameworks. New York City's Local Law 144 requires an outcome-based bias audit regardless of intent. TRAIGA does not. The statute is built around what the deployer meant to do, not what the tool happened to produce.

What proving intent actually requires

Intent-based liability under TRAIGA is harder to establish than outcome-based liability, which is precisely the point of the design. A regulator or claimant has to show that a company developed or deployed a system with the purpose of producing discriminatory treatment, not merely that a disparity resulted.

In practice, this generally requires something closer to direct evidence: internal communications indicating discriminatory purpose, design choices that cannot be explained except by reference to a protected characteristic, or a documented decision to ignore known discriminatory effects while continuing to deploy the tool unchanged. A tool that simply reflects historical hiring patterns, without evidence anyone intended that outcome, is a much harder case to bring under section 552.056(c) than the same tool would be under an outcome-based statute.

Why this protection is narrower than it looks

The intent standard protects against TRAIGA liability specifically. It does not touch federal law. Title VII, the Age Discrimination in Employment Act, and the Americans with Disabilities Act all permit claims based on disparate impact without any showing of intent. A hiring tool can be fully compliant with TRAIGA's intent standard, having never been designed or deployed with discriminatory purpose, and still create real exposure under federal law if it produces a disparate outcome that cannot be justified as job-related and consistent with business necessity.

This means a Texas employer cannot treat TRAIGA compliance as a substitute for adverse impact analysis. The two regimes test different things, and passing one says nothing about the other.

Courts and agencies still look at impact when assessing intent

Even within TRAIGA's own framework, disparate impact is not irrelevant. It is simply insufficient on its own. Impact evidence is frequently how intent gets established in practice, both under TRAIGA and more generally in discrimination law. A disparity that is discovered, documented internally, and then left uncorrected across repeated hiring cycles starts to look less like an accident and more like a choice, which is exactly the kind of evidence that can support an intent finding.

The practical lesson: running adverse impact analysis is not something the intent standard makes optional. It is precisely the analysis that determines whether an employer can show it acted in good faith once a disparity is found, which matters both for TRAIGA and for federal exposure.

What this means for a Texas hiring stack

1. Do not read the intent standard as license to skip adverse impact analysis. Federal law does not share TRAIGA's threshold, and skipping this analysis leaves the larger exposure uncovered.
2. Document your design and deployment decisions, so that if a disparity is later found, the record shows the tool was not designed or knowingly deployed to produce it.
3. Treat a discovered disparity as something to fix, not just monitor. Leaving a known disparity uncorrected is the pattern most likely to convert an outcome issue into an intent issue over time.
4. Remember TRAIGA's safe harbor operates independently of the intent standard. Documented alignment with a recognized AI risk management framework is a separate defense worth building regardless of how strong the intent protection feels.

Frequently asked questions

Does TRAIGA require a bias audit like Local Law 144 does? No. TRAIGA imposes no audit requirement on private employers. The intent standard means outcome measurement is not a statutory obligation, though it remains good practice given federal exposure.

If our tool shows a disparate impact, are we automatically protected under TRAIGA? The disparity alone does not establish a TRAIGA violation. But a documented, known disparity left uncorrected can become evidence supporting an intent finding over time, and it remains fully actionable under federal law regardless of TRAIGA.

Does the intent standard apply to disability, age, and other protected classes, or just race and sex? TRAIGA's prohibition covers intent to discriminate against a protected class under federal or state law generally, not a narrower list.

Is this different from how TRAIGA treats government use of AI? Yes. TRAIGA places more extensive obligations on government agencies than on private employers, whose obligations are comparatively narrow and centered on this intent-based prohibition.

Should we still run adverse impact analysis if we only worry about TRAIGA? Yes. Federal law does not share TRAIGA's intent threshold, and adverse impact analysis is the practical tool for managing that separate, larger exposure.