WCAG 3.2 Draft Expands to Cover AI-generated Content and Live Captioning Standards

Tech · 6 min read

Key changes in the draft address provenance and clarity of AI-generated content: systems must provide meaningful alternative text when images are synthesized, clarify when content is generated vs. authored, and ensure automated transcripts meet minimum accuracy thresholds before being used as sole access mechanisms.

For live captioning and transcription, the draft outlines measurable quality criteria (word-error rate targets, speaker labeling, and latency limits) and requires fallback options if automated captions fall below thresholds. The intent is to drive procurement and engineering practices toward high-quality captioning rather than treating 'some captioning' as sufficient.

The draft is advisory for now, but major platform vendors and accessibility consultancies are already recommending that product teams treat these criteria as de facto standards. Design and engineering teams should audit AI workflows and live-media pipelines to identify compliance risks ahead of finalization.