{"slug": "why-whisper-and-medical-speech-apis-are-making-potentially-fatal-errors", "title": "Why Whisper and medical speech APIs are making potentially fatal errors", "summary": "Appen published an analysis warning that OpenAI's Whisper and medical speech APIs are producing transcription errors that can be potentially fatal in clinical settings. Appen cites its own 30 years of AI data expertise, SOC 2 and ISO 27001 certifications, and coverage of 500+ global locales as the basis for its model evaluation work on hallucination benchmarking, bias, and compliance audits.", "body_md": "Data products\nData products\nFrontier Alignment\nCoT reasoning, SME RLHF, SFT & red teaming\nAgentic AI\nAgent trajectories, RL environments & tool-use logs\nSpeech & Audio\nExpressive TTS, emotion detection & dialectal speech\nMultimodal AI\nVLM data, video annotation & cross-modal alignment\nPhysical AI\nLiDAR annotation, robotics trajectories & sensor fusion\nModel Integrity\nHallucination benchmarking, bias & compliance audits\nWhy Appen\n30 years of AI data expertise\nSOC 2 & ISO 27001 certified\n500+ global locales covered\nIndependent model evaluation\nOTS datasets\nEnterprise data\nPerspectives\nContent\nBlog\nExpert commentary on frontier AI development\nResearch\nResearch papers and publications from Appen\nCase Studies\nHow leading AI labs work with Appen\nPodcasts\nExpert conversations on AI and data\nWebinars & Events\nLive and on-demand sessions with our experts\nFeatured\nWhy Human Evaluation Still Wins\nMaster RLVR at Scale\nImproving LLM Safety Evaluations\nAbout\nCompany\nAnnotation Platform\nAI Data Platform (ADAP) — automation meets human oversight\nAbout Us\n30 years pioneering AI data\nInvestor Relations\nASX-listed: APX\nNews\nLatest announcements and press\nCareers\nBuild the future of AI with us\nGet in touch\nJoin our workforce\nJoin our workforce\nGet in touch", "url": "https://wpnews.pro/news/why-whisper-and-medical-speech-apis-are-making-potentially-fatal-errors", "canonical_source": "https://www.appen.com/appen-medterm-90-benchmark", "published_at": "2026-09-17 22:23:50+00:00", "updated_at": "2026-09-17 22:55:26.178214+00:00", "lang": "en", "topics": ["ai-safety", "natural-language-processing", "ai-ethics"], "entities": ["Appen", "Whisper", "OpenAI"], "alternates": {"html": "https://wpnews.pro/news/why-whisper-and-medical-speech-apis-are-making-potentially-fatal-errors", "markdown": "https://wpnews.pro/news/why-whisper-and-medical-speech-apis-are-making-potentially-fatal-errors.md", "text": "https://wpnews.pro/news/why-whisper-and-medical-speech-apis-are-making-potentially-fatal-errors.txt", "jsonld": "https://wpnews.pro/news/why-whisper-and-medical-speech-apis-are-making-potentially-fatal-errors.jsonld"}}