company

UnitedHealth Group

Managed Healthcare & Insurance
25Poor
Aio Composite Score
Confidence
ApplicationHeavy
high
ImpactMixed
high
OpennessMisleading
high
UnitedHealth Group is one of the most aggressive AI adopters in healthcare, deploying over 1,000 AI applications across its insurance, care delivery, and pharmacy divisions, including AI chatbots that handled 65 million customer calls in 2024 and 20,000 engineers generating AI-assisted software code. However, the company faces a serious class-action lawsuit alleging its nH Predict AI tool denied Medicare Advantage claims with a 90% error rate, and its public claims that AI is never used for coverage decisions are directly contradicted by this lawsuit and a U.S. Senate investigation. Significant layoffs in 2024-2025 have been reported, though a clear causal link to AI displacement specifically remains circumstantial rather than evidenced. The gap between UnitedHealth's 'responsible AI' public narrative and the documented harms from nH Predict raises material integrity concerns.
Application
How is AI being used?
Heavy
high
UnitedHealth has 1,000+ AI applications in production across claims, call centers, clinical support, software engineering, and pharmacy — confirmed…
Impact
How have people been affected?
Mixed
high
There is direct evidence of AI-driven customer harm: the nH Predict lawsuit alleges a 90% error rate in denying Medicare Advantage claims,…
Openness
Company honesty about AI use.
Misleading
high
UnitedHealth's public FAQ categorically states AI is never used for coverage, prior authorization, or claims decisions — but a federal class-action…
AI Tools & Platforms Detected
nH Predict (NaviHealth)AVA (Agent Virtual Assistant)United AI Studio (UAIS)Smart Choice (provider search)Advocate AssistOptum Real (real-time claims AI)Azure OpenAIHugging FaceLevel2 (digital therapy platform)Optum AI
AI Hiringmoderate
30AI roles/ 7144 total
Principal AI/ML EngineerSenior Director, AI/ML EngineeringAI/ML Engineer (OptumRx)AI/ML Research Intern - PhD LevelPrincipal AI/ML Software Engineer
Key Evidence (10)
UnitedHealth has over 1,000 AI applications in production, including AI chatbots handling 65 million customer calls in 2024 and 18 million AI-enabled doctor searches in Q1 2025 alone; 20,000 engineers use AI tools generating 60 million+ lines of code
2025-05-28
web searchSource ↗
A class-action lawsuit alleges UnitedHealth's nH Predict AI tool denied Medicare Advantage claims with a 90% error rate, overriding doctors' judgments; a STAT News investigation cited in the suit claims employees were pressured to keep patient rehab stays within 1% of nH Predict's predictions
2024-12-09
web searchSource ↗
UnitedHealth's public FAQ explicitly states 'No' to AI being used for coverage decisions, prior authorization decisions, clinical decisions, and claims rejection — directly contradicted by the nH Predict lawsuit and U.S. Senate Permanent Subcommittee on Investigations report
web searchSource ↗
UnitedHealth's 2024 10-K identifies AI/ML including generative AI as a material risk factor, noting failure could cause regulatory sanctions, fraud vulnerabilities, and adverse business outcomes
2025-02-27
sec filingSource ↗
UnitedHealth's 2023 10-K (filed 2024) also identifies AI/ML as a risk factor with near-identical language, indicating this has been a disclosed material risk across consecutive annual filings
2024-02-28
sec filingSource ↗
Job postings for Senior Director AI/ML Engineering, Principal AI/ML Engineer, AI/ML Research Intern (PhD), and Senior AI/ML Engineer at OptumRx reference United AI Studio (UAIS), Machine Learning Review Board (MLRB), Azure OpenAI, Hugging Face, LLM integration, and fraud detection models
2026-01-28
job postingSource ↗
What insider reports could clarify
  • Has nH Predict been decommissioned, modified, or is it still in active use for patient decisions?
  • Are benefits unit layoffs being backfilled with AI tools rather than rehires?
  • How does the AI Review Board actually override or reject a deployed AI model in practice?
  • Do call center employees have meaningful ability to override AI-generated recommendations?
  • Have patients been notified when AI systems influenced decisions about their care?
Submit a ReportClaim This Profile