| AI doomsday predictions could put the brakes on health tech innovation. |
| At least one healthcare company — electronic health record giant Epic, which offers AI tools for hospitals and clinicians — has paused tech development for at least six weeks to ensure its systems are secure, Modern Healthcare reported. The move comes after rogue AI agents at OpenAI, Anthropic and elsewhere hacked systems all on their own. |
| Generative AI, of course, has made big inroads into healthcare over the last couple of years. Hospital systems first adopted ambient scribes, and then turned to tools for prior authorizations, coding and billing, and clinical decision support. Now the federal government is encouraging the development of AI for
medical care delivery, and startups are meeting the challenge. |
| But as I wrote Tuesday, it’s possible the public’s growing distrust of the technology could derail that effort. Leaders of OpenAI and Anthropic seem to agree that the pace of AI development should slow down. |
| Robert Wachter, professor of medicine at UCSF, wrote in his blog last week that slowing AI in healthcare would be a mistake, however. Putting the development of frontier models on ice is sensible; halting healthcare’s adoption
of models already available could cost lives, he said. |
| “The peculiar fact about medicine is that we don’t need the frontier labs to keep pushing our frontier,” Wachter wrote. “The American healthcare system is a mess, failing to meet our needs in quality, safety, access, equity and cost. This means that the AI we already have is good enough to address many of our most important clinical problems.” |
| It seems his fear might be materializing. Epic’s CEO Judy Faulkner said that going forward, the company will develop at a different pace, because of the constant need to protect against cyberattacks. |
| Will others follow Epic’s lead? |
| - Shelby |
