ChatGPT's health push: 230M weekly users and a new research lead
OpenAI researcher Karan Singhal is leading an ambitious effort to make ChatGPT a trusted healthcare tool, as more than 230 million people already use it for health and wellness advice each week.
Score breakdown
OpenAI's systematic investment in health-specific training, physician partnerships, and a dedicated evaluation benchmark marks a concrete escalation of ChatGPT's role in medical guidance at a scale — 230 million weekly users — that already rivals major consumer health information platforms.
- 01More than 230 million people use ChatGPT for health and wellness advice each week, according to OpenAI.
- 02Karan Singhal, who previously developed Google's Med-PaLM medical AI models, leads OpenAI's health research efforts.
- 03OpenAI's GPT-5 model family is the first trained specifically at every stage of development to improve health advice.
Karan Singhal, a researcher who previously helped develop Google's Med-PaLM medical AI models, joined OpenAI in mid-2024 and has since built a dedicated health research team aimed at making ChatGPT a reliable source of medical guidance. According to the Business Insider article by Stephen Council, more than 230 million people use ChatGPT for health and wellness advice each week — a figure OpenAI attributes in part to Singhal's work. Singhal's stated goal is to make ChatGPT good enough at health that it meaningfully improves people's lives, avoids harm, and wins over skeptics, including a shift he says is already underway in which patients view OpenAI's models as a "protector in their care journey."
Singhal also noted that he sees doctors rapidly adopting ChatGPT for Clinicians and other AI tools, and characterized healthcare's adoption of technology as "incredibly, insanely fast."
OpenAI's GPT-5 model family is described as the company's first to be trained specifically at every stage of development to be better at health advice. To support that effort, Singhal partnered with more than 200 physicians — framing it as "aggregating the wisdom of the crowd" — and helped launch HealthBench, a physician-developed evaluation suite for measuring AI health capabilities. OpenAI reported that GPT-5.5 Instant scored better than both physician-written answers and GPT-4o in HealthBench tests, and that comparing billions of anonymized health-related messages revealed a 71% drop in flagged inaccurate responses over the past two months. The article notes that OpenAI has faced lawsuits alleging GPT-4o encouraged suicidal ideation and gave harmful advice — claims OpenAI has denied — but the company has continued to deepen its health focus rather than pull back. Singhal also noted that he sees doctors rapidly adopting ChatGPT for Clinicians and other AI tools, and characterized healthcare's adoption of technology as "incredibly, insanely fast."
Key facts
- 01More than 230 million people use ChatGPT for health and wellness advice each week, according to OpenAI.
- 02Karan Singhal, who previously developed Google's Med-PaLM medical AI models, leads OpenAI's health research efforts.
- 03OpenAI's GPT-5 model family is the first trained specifically at every stage of development to improve health advice.
- 04Singhal built a team of health researchers and established partnerships with more than 200 physicians.
- 05OpenAI helped launch HealthBench, an evaluation suite created with physicians to measure AI health capabilities.
- 06GPT-5.5 Instant scored better than both physician-written answers and GPT-4o in HealthBench tests.
- 07OpenAI reported a 71% drop over the past two months in health responses flagged for inaccuracy.
Topics
Summary and scoring are generated automatically from the original article. We always link back to the publisher and never republish images or paywalled content. Last processed Jun 20, 2026 · 08:55 UTC. How this works →