HHS eases AI translation oversight as California bill seeks to ban unreviewed AI in healthcare
Federal regulators loosened AI translation rules in healthcare, while California’s pending AB 1242 would require human review. Learn what this means for providers and patients.

TL;DR
In May 2024 HHS OCR rescinded the requirement that AI‑generated translations be checked by a human, and in December 2024 allowed delayed review only in emergencies. Meanwhile, California’s pending AB 1242 would define AI translation without human review as not qualifying as a certified interpreter.
Context: Federal rules under Section 1557 of the Affordable Care Act require qualified interpreters for patients with limited English proficiency, but the regulations do not explicitly address machine translation. As AI tools become more common, regulators have struggled to clarify when human oversight is necessary.
Regulations detail: The implementing regulations under Section 1557 call for a qualified interpreter who demonstrates proficiency in English and another language and follows accepted ethics principles, including confidentiality. These criteria were written before generative AI became widely available, leaving room for interpretation about whether AI output can meet the standard.
Key Fact 1: In May 2024 the HHS Office for Civil Rights announced it had removed parts of the 2020 Section 1557 rule that mandated human review or patient notification for machine‑translated health documents.
Key Fact 2: In December 2024 HHS OCR issued a Dear Colleague Letter permitting delayed human review of AI translations during emergencies, provided the review occurs as soon as practicable and patients receive a warning about possible errors.
Key Fact 3: California’s AB 1242, currently under Senate consideration, would exclude AI translation that lacks human review from the definitions of “qualified interpreter,” “qualified translator,” and “translation."
What It Means: Providers relying on AI translation must now navigate a split landscape: federal guidance permits limited, emergency‑only use without immediate human checks, while California law would bar such use entirely unless a certified interpreter verifies the output. Practical takeaway: healthcare organizations operating in California should retain human review for all AI‑generated translations to stay compliant, whereas those in other states may adopt emergency exceptions but should still warn patients of potential inaccuracies. No clinical trial data were cited in the announcements, so real‑world error rates remain uncertain.
Watch for the Senate vote on AB 1242 later this year and any further HHS guidance that could clarify emergency‑use criteria.
Continue reading
More in this thread
Santa Clara County’s TRUST Program Faces Funding Cuts Despite 90% Stabilization Rate
Dr. Priya Sharma
HHS Scales Back AI Translation Safeguards as California Moves to Ban Unreviewed AI in Care
Dr. Priya Sharma
SSM Health Builds Oklahoma’s First Hospital Solar Field to Cut Emissions Half by 2030
Dr. Priya Sharma
Conversation
Reader notes
Loading comments...