AI’s dual impact
“The integration of artificial intelligence and machine learning in health care has a dual impact on cybersecurity,” according to Michael Haas, a health care senior analyst at RSM US LLP.
“On one hand, these technologies enhance security by providing checks and balances, identifying anomalies, and ensuring that patient messages and administrative tasks are properly vetted," he says. "AI helps organizations move from merely performing tasks to adopting an auditor role, thereby improving oversight and security.”
On the other hand, he says, AI and machine learning can also introduce new vulnerabilities. Many organizations are unaware of the AI components within their systems, which can lead to unintentional exposure to cyberthreats. Haas stresses the importance of creating a robust AI governance structure to manage these technologies effectively.
Regulatory bodies such as the Office for Civil Rights, National Institute of Standards and Technology (NIST), Centers for Medicare and Medicaid Services, and U.S. Department of Justice are imposing stricter guidelines and frameworks to ensure health care organizations maintain robust cybersecurity measures. Haas cites the NIST 2.0 framework and the Trusted Exchange Framework and Common Agreement as examples of efforts to enhance cybersecurity and improve the nationwide exchange of electronic health information across different health care entities. While these frameworks are voluntary, their implementation is strongly encouraged to reduce cyber vulnerabilities and meet eligibility requirements for cyber insurance.
Collaborations with third-party vendors pose additional cybersecurity challenges, says Haas.
He highlights the importance of selecting vendors with appropriate accreditations, such as the Health Information Trust Alliance (HITRUST) certification, to ensure they meet stringent security standards. Liability shifts are also becoming more common, with organizations demanding higher levels of liability, credentials and accountability from vendors to cover potential data breaches.