How Safe Is AI in Healthcare?

So I haven't talked about AI recently and the Center for Connected Health Policy, which is also the National Telehealth Policy Resource Center, they put out a policy called Push and Pull, and we're, we have been talking about the recent federal and state AI activity, and I think it's important that we take a look at this.

This was published on June 23rd, and I just, y-you know, it's interesting because we have so many federal and then now state laws passing trying to regulate artificial intelligence systems, and we need to talk about this because it's, again, coming at us like a freight train.

So as states continue to pass laws regulating the development and the deployment of AI, the federal government is also ramping up its efforts to create a national AI regulatory framework, potentially overriding some of the state AI laws.

So the recent federal policy proposals and discussions were consistent with the, uh, president's December 25th executive order talking about ensuring a national policy framework for AI, and it was in the January 7th newsletter of, uh, the CCP.

But according to the executive order, which I read, the order is intended to preserve US global leadership in AI by reducing regulatory barriers, addressing what administration characterizes as fragmented state-level regulatory landscape, and advancing more of a uniform national AI policy framework.

The problem I see is that it doesn't always specifically mention healthcare, and we really need, you know, to, to look at healthcare as a rule here because as I am continuing my efforts fr-from an auditing perspective, both on the physician and payer side, I'm seeing ambient scribes, I'm seeing AI, I'm seeing hallucinations and just so many errors and just not checking and making sure that, you know, you're verifying what is actually in the note.

Uh, earlier this month, another executive order called Promoting Advanced Artificial Intelligence, Innovation and Security was issued, but this EO was focus, uh, focused, I'm sorry, around prioritizing the cyber defense of national security systems, which is important also in healthcare because that, you know, looks at developing advanced AI vurnal-- vulnerability detection, including facilitating access to cybersecurity tools and looking at infrastructure, especially at hospitals, especially rural hospitals.

And it also speaks to, you know, securing, um, a, a really, the most high-risk and advanced AI models development and testing this to make sure that, again, it's not biased, it's not hallucinating, it, it can't be hacked, you know, on a regular basis.

This is just very, um, crazy to me just really looking this because they're looking at, uh, uh, Anthropic, they're looking at OpenAI, they're looking at ChatGPT, they're looking at a lot of different things that, you know, public release enabling early evaluation and mitigation of potential risks, but they really have to bring it back also and realize that there's certain niches here, like healthcare, that these advanced models, um, also third-party auditors and things that you see off country that are gonna be a problem.