Updated
Updated · The New York Times · Sep 30
Epic Systems Uses AI to Patch Flaws Threatening 325 Million Patient Records
Updated
Updated · The New York Times · Sep 30

Epic Systems Uses AI to Patch Flaws Threatening 325 Million Patient Records

3 articles · Updated · The New York Times · Sep 30

Summary

  • Epic Systems said it is deploying AI tools to find and fix security weaknesses that could let hackers gain undetectable access to patient health data.
  • Anthropic’s Claude Mythos is being used to stress-test Epic’s software and probe how attackers might use open-source AI agents to unlock confidential records.
  • The urgency pushed Epic to slow development on some product lines while engineers spent six weeks focused largely on closing the gaps, CEO Judy Faulkner said.
  • The threat is especially significant because Epic is the largest U.S. medical-records vendor, serving thousands of hospitals and doctors’ offices and holding records for 325 million patients worldwide.

Insights

Epic used Anthropic’s AI to hunt a hidden flaw in records for 325 million patients—but what exactly did the system find?
If Epic had to slow product work for six weeks to fix one unusual weakness, how secure are today’s AI-powered health platforms really?