Editing News-July-18-2026

'''July 18, 2026''' — The United Kingdom's AI Security Institute reported that recent open-weight AI models were narrowing the cyber-capability gap with frontier closed models, while U.S. Medicare officials' WISeR pilot highlighted the regulatory stakes of using AI in coverage review.<ref name="aisi-open-weight-cyber">[https://www.aisi.gov.uk/blog/how-far-behind-the-frontier-are-leading-open-weight-models-on-cyber AI Security Institute, "How Far Behind the Frontier are Leading Open Weight Models on Cyber?" (July 18, 2026)]</ref><ref name="ars-medicare-ai-prior-auth">[https://arstechnica.com/ai/2026/07/will-ai-fix-prior-authorization-or-make-it-worse/ Ars Technica / Undark, "Will AI fix prior authorization—or make it worse?" (July 18, 2026)]</ref><ref name="cms-wiser">[https://www.cms.gov/priorities/innovation/innovation-models/wiser Centers for Medicare & Medicaid Services, "WISeR (Wasteful and Inappropriate Service Reduction) Model" (accessed July 19, 2026)]</ref>

== Contents ==
# [[#UK AI Security Institute says open-weight cyber capabilities are closing on frontier models|UK AI Security Institute says open-weight cyber capabilities are closing on frontier models]]
# [[#CMS prior-authorization pilot puts AI coverage review in focus|CMS prior-authorization pilot puts AI coverage review in focus]]

----

== UK AI Security Institute says open-weight cyber capabilities are closing on frontier models ==
The AI Security Institute said it has tracked frontier-model cyber capabilities since 2023 and found that the most capable open-weight models it evaluated still lagged leading closed frontier systems.<ref name="aisi-open-weight-cyber" /> The institute reported that recent open-weight models lagged the cyber capabilities of frontier closed models by roughly four to seven months, compared with a six-to-ten-month gap through most of 2025.<ref name="aisi-open-weight-cyber" /> The institute said its open-weight-model evaluations were largely unimpeded by safeguards and that DeepSeek V4-Pro refusals on narrow cyber tasks were easy to circumvent in testing.<ref name="aisi-open-weight-cyber" /> The findings matter for AI governance because open-weight releases can distribute frontier-adjacent cyber capability beyond a single hosted provider's access controls.<ref name="aisi-open-weight-cyber" />

----

== CMS prior-authorization pilot puts AI coverage review in focus ==
Ars Technica reported that the Centers for Medicare & Medicaid Services began the WISeR demonstration project as the Trump administration explored using AI in Medicare coverage-review workflows.<ref name="ars-medicare-ai-prior-auth" /> CMS describes WISeR as a Wasteful and Inappropriate Service Reduction Model that will test technology-assisted prior authorization for selected Original Medicare services in six states.<ref name="cms-wiser" /> Ars Technica reported that the model combines machine-learning tools with human clinical review to evaluate services that CMS considers vulnerable to overuse, fraud, and abuse.<ref name="ars-medicare-ai-prior-auth" /> The pilot is relevant to AI law and policy because coverage-review automation can affect access to care, administrative due process, and oversight of high-stakes healthcare AI systems.<ref name="ars-medicare-ai-prior-auth" /><ref name="cms-wiser" />

== References ==
<references />

[[Category:International]]
[[Category:United Kingdom]]
[[Category:Transparency]]
[[Category:Federal Regulation]]
[[Category:Healthcare AI]]
[[Category:Sector-Specific Regulation]]