Nine U.S. health systems now say patient information linked to Xsolis, a healthcare AI vendor, was exposed in a phishing attack that began in January. That is up from seven in earlier coverage, and the tally now puts about 1.4 million patients in the affected pool. The breach is not new. Xsolis first disclosed it in June, but the list of affected hospitals, which includes Mayo Clinic and UW Medicine, is only now coming into focus.

Xsolis sells utilization management and care coordination software to hospitals and health plans. Patient records pass through its systems as part of routine decisions about care, which is why an intrusion at a vendor like this one reaches far beyond the company itself. The story also matters because hospitals increasingly send patient data to third-party AI tools.

What Xsolis disclosed and when

The timeline is short. According to a Paubox summary of the notice, a targeted phishing attack on January 20, 2026 gave an unauthorized actor access to part of the Xsolis environment. The company detected the activity two days later. TechTimes gives the detection date as January 22, which matches that account.

Xsolis reported the incident in a June 5 filing with the U.S. Department of Health and Human Services. In that notice, the company said the intruder took a "limited number" of files. The company has not described that number further, and the sources do not say how many files were involved or what they held.

HHS posted the 1.4 million figure on June 22, based on what Xsolis told the agency. TechTimes counted 1,396,519 individuals from the breach notices, which rounds to the same total. Xsolis says it is not aware of any misuse of the data and is notifying affected patients.

Which health systems are on the list

The nine health systems now reported as affected are:

  • Erie County Medical Center
  • Mayo Clinic
  • UW Medicine
  • Legacy Health
  • VHC Health
  • Rochester Regional Health
  • Carle Health
  • Augusta Health
  • Hendrick Health

Mayo Clinic and UW Medicine are the best-known names here, but several others are regional systems. Rochester Regional Health said about 18,600 of its patients were affected. It also pointed out that the breach involved a vendor it no longer works with.

Legacy Health gave patients a more specific account. Its affected records may include medical record numbers, diagnoses, treatment dates and provider names. The health system said those records do not include Social Security numbers or financial information.

What data may have been exposed

Xsolis says the exposed information varies by person. It may include names, addresses, dates of birth, health insurance information, Social Security numbers and medical treatment information. That range matters. A patient whose file held only a name and a treatment date faces a different level of risk than one whose file included a Social Security number and insurance details.

The sources do not say which patients at each health system had Social Security numbers exposed, and that is the detail most people will want. Legacy Health's notice rules out those numbers in its own affected records, but it does not settle the question for the other eight systems. Patients will learn their status mainly through notification letters, which Xsolis says it is sending.

Why the count of affected systems keeps changing

Earlier coverage did not agree on how many providers were involved. Becker's reported seven health systems earlier this year, and DistilINFO listed eight. The newest tally, from Becker's and DataBreaches.net, lists nine.

The sources I reviewed do not explain the gap. Readers should treat nine as the current count, not a final one, because more providers could still confirm involvement.

Why an AI vendor breach matters for hospitals

Xsolis belongs to a growing group of vendors that sit between hospitals and patient records. Hospitals hand data to outside software to support coverage decisions and to coordinate care. Every vendor in that chain is another place an attacker can try to get in. In this case, the way in was ordinary: a phishing attack, which the sources describe as a common entry point for breaches at third-party vendors.

One outlet called this breach among the most significant healthcare vendor cyberattacks of 2026. The 1.4 million total explains that concern. It is nearly three times the size of a 2025 case that Becker's has compared it to.

In May 2025, Serviceaide disclosed that a Buffalo, N.Y. health database had been publicly exposed, affecting 483,126 patients. That was a different kind of failure, an exposed database rather than a phishing break-in. Both cases show how much sensitive health information can sit in a single vendor's systems.

Data exposure is not limited to health records. Revolve's data breach exposed saved payment cards for up to 1.5 million people, which shows how the type of data at stake depends on who is holding it.

What patients should do now

Several steps can help patients protect themselves:

  • Check whether you received care from any of the nine health systems listed above. If you did, you may be among the patients the notices cover.
  • Watch for breach-notification letters, and verify any letter before acting on it. Call the health system using a number from its official website, not one printed in the letter.
  • If your Social Security number may be involved, place a free credit freeze with Equifax, Experian and TransUnion. A freeze blocks most new credit checks until you lift it.
  • Treat unexpected calls or emails that mention your care or insurance as possible phishing. Hang up and call your provider back on a number you have verified.
  • Ask your provider which vendors hold your records, and whether Xsolis is one of them.

What comes next

Several open questions could change the picture in the coming months. Xsolis has said nothing beyond its June 5 statement, so any fuller account of how many files were taken and what they contained would be new information. The nine-system count may grow, since the reporting disagrees and some systems may still be confirming involvement.

The legal action that Becker's reports has already been filed could produce court filings that describe the claims and name the parties. For hospitals, the test is whether they begin asking AI and data vendors for specific proof of how they guard access, rather than general assurances. For patients, the practical question is simpler: whether a letter arrives, and whether it says a Social Security number was involved.