The preprint by Caroline S. Wagner tests whether research security policies have redirected international collaboration away from sensitive fields, using eight years of bibliometric data on China's output across 27 Scopus categories from 2018 to 2025. Contrary to deterrence assumptions, declines in international co-authorship were smallest (3-6 percentage points) in high dual-use areas such as computer science, materials science, physics, and engineering, while non-sensitive fields like psychology and nursing saw steeper drops (8-24 points). This pattern indicates researchers adapted forms of engagement rather than withdrawing from frontier work. As a systems-level preprint, not yet peer-reviewed, the study relies on aggregate publication shares without granular author-level or institutional controls, limiting causal claims about specific policies like US export controls or China's domestic capacity push. Related analyses, including a 2023 Science Policy Forum piece on CHIPS Act impacts and a 2024 Research Policy study of EU-US decoupling trends, reveal the same adaptive pattern: networks persist through domestic intermediaries and language adjustments. These under-reported dynamics expose systemic tensions where national security framing overrides open-science norms, echoing governance shifts seen in semiconductor export rules and visa scrutiny that prioritize selective decoupling over broad suppression.