China champions open AI to the world. In private, it’s afraid of one closed American model.
THE SO WHAT
Beijing’s reported anxiety over Anthropic’s closed Mythos model as a potential cyber weapon underscores how frontier AI is now viewed as dual-use infrastructure. Multinationals operating across US–China lines should expect AI export controls and model access regimes to tighten and plan architectures that can tolerate that fragmentation.
READ THE SOURCE
MORE FROM THE WIRE
Applied AIAI coding agents are blowing through budgets — Replit, Kilo Code, and Symbotic explain how they're managing it
When engineers only touch code 1% of the time, your cost center shifts from people to tokens — and ungoverned agents will happily set fire to your cloud bill. Treat coding agents like any other high-throughput service: enforce budgets, define which repos and systems they can touch, and make review workflows explicit before scaling usage.
Applied AIAnthropic names Mariano-Florentino Cuéllar, an ex-California Supreme Court justice and a special assistant in Obama's WH, as its first global affairs chief
Top-tier legal and policy talent moving in-house at labs is a sign that AI governance is becoming a direct executive lane, not a side function. If your product depends on frontier models, assume the regulatory conversation will increasingly be lab-to-state — and plan your own policy, compliance, and lobbying capacity accordingly.
Applied AII tried to use ChatGPT to create fake evidence — and I came away more worried than I expected
The gap between what models refuse and what they still enable is where real reputational and legal exposure lives. If your workflows touch identity, compliance, or disputes, assume generative tools will be used to fabricate artifacts and invest now in provenance checks, cross-system verification, and staff training on synthetic media.
Applied AIOpenAI pays $3.2m to settle a US claim it favoured visa workers over Americans
A $3.2M settlement over alleged hiring bias toward visa workers puts AI talent strategy squarely in the crosshairs of labor regulators. If you’re scaling technical teams in the US, tighten your PERM and recruiting processes now — and be ready to evidence non-discriminatory practices under scrutiny.