SHRM26 Debrief · Legal & Compliant HR · #2035

Strategies for Navigating ADA Compliance in the Age of Artificial Intelligence

with Trish Zuniga

▶ Watch on the SHRM portal ~44 min, distilled

ADA compliancereasonable accommodationAI governance

"AI systems don't pause. They don't interpret nuance, and they don't really stop and think about what your employees actually need. They just execute."

What it was about

AI tools embedded across the reasonable accommodation lifecycle can silently break the ADA's requirement for an interactive, individualized, human process, so employers must anchor AI governance in responsible-AI principles (mapped via frameworks like NIST) rather than chasing constantly-changing AI regulations.

By the numbers

up to 4 times more

IBM study finding on how much more of their AI budget employers spend if they retrofit AI governance instead of building it in from the outset

Key notes

Map every point where AI touches the accommodation lifecycle (intake, information gathering, interactive process, evaluating options, implementation, monitoring, documentation) and decide human-in-the-loop vs. human-on-the-loop for each step.
Assign explicit accountability for AI/ADA compliance across HR, IT, legal, and vendors so no one can claim 'someone else owned it' if a dispute arises.
Apply the NIST AI Risk Management Framework's Govern-Map-Measure-Manage cycle as a stable, law-agnostic foundation instead of reacting to each new AI regulation.

The contrarian takeThe ADA itself doesn't actually require documenting the interactive process, but courts and internal counsel treat undocumented accommodation efforts as if they never happened, so documentation becomes a practical necessity beyond the letter of the law.

Take this back Monday

Do this for your team

Map every AI tool touching accommodation requests (intake to documentation) and flag where a human, not the AI, must make the call.

Say this in your next leadership meeting

AI can process accommodation requests, but it can't run the ADA's interactive process for us — that still has to be a human conversation.

Watch out for

Treating AI governance as an abstract compliance exercise disconnected from real workflow breakpoints, like when a hiring manager blindly trusts an unexplainable AI score.
Letting AI-enabled workflows execute automatically without pausing for the human judgment the ADA's interactive process legally requires.
Chasing individual, rapidly-changing AI laws (EU AI Act, Colorado's law) jurisdiction by jurisdiction instead of building on stable responsible-AI principles.

Fun fact · Trish Zuniga

She holds two law degrees — one from George Washington University and one from Ateneo de Manila University in the Philippines.

If this landed, see these

↳ Go deeperSeven Steps for Perfecting Your DocumentationTurns 'document the interactive process anyway' into a full seven-step documentation discipline.⇄ The counterpointEthical AI to Broaden Talent Pools and Reduce Hiring BiasReframes the risk: candidates vanish from AI pipelines before any interactive process even starts.✦ The unexpected oneBuilding Culture by Design: Stop Leaving Your Greatest Asset to ChanceBoth break the same way: silently, by default, unless someone designs them on purpose.