Four Z.ai founders participated in a 2018 to 2021 research project co-funded by CGTRI, which the UK's MI5 said had "very strong ties" to China's main spy agency
THE SO WHAT
AI talent and research lineages are now being scrutinized like supply chains—who funded your early work is becoming a compliance and trust question. Multinationals partnering with frontier labs need a playbook for provenance checks and geopolitical exposure, not just model evals.
READ THE SOURCE
MORE FROM THE WIRE
Applied AIPartnering with Blacksmith
If agents can write production-scale code, the new bottleneck is proving that code is safe, performant, and compliant at the same speed. Expect “agent-era CI/CD” vendors like Blacksmith to become a required layer in any serious agent deployment, not a nice-to-have devtool.
Applied AISpotify pulled 50 million AI slop and spam tracks from its recommendations
Fifty million AI spam tracks getting pulled means recommendation systems are now content moderation systems at scale. Any platform that allows generative uploads needs an explicit anti-spam and provenance strategy or its core UX will quietly degrade.
Applied AIFigma co-founder Dylan Field thinks MCP is not a perfect protocol
Figma signaling that MCP isn’t “perfect” while promising to support multiple standards is a reminder that no one wants a single agent integration choke point. If you’re betting on MCP, design your extensions so they can be exposed via at least one other protocol without a full rewrite.
Applied AIPangram review: A polished AI detector that gets the basics right
Tools like Pangram that are strong on fully AI-generated text but weaker on hybrids underscore how brittle AI detection still is. If your compliance model assumes you can reliably detect AI content, you’re carrying more risk than your policies admit.