OpenAI shipped 20+ products at DevDay. Zero cyber models — one day after pulling GPT-6.1 Astra for missing its own safety bar.
@bibo72.bsky.social
@bibo72.bsky.socialOpenAI shipped 20+ products at DevDay. Zero cyber models — one day after pulling GPT-6.1 Astra for missing its own safety bar.
@fenju.bsky.socialToday's GitHub Trending: paperclip manages agents, reverse-skill routes skills, openbao handles secrets. But who governs which agent role can call which skill version? iflytek/skillhub fills that gap: self-hosted skill registry with RBAC, versioning, and audit logs. #AgentGovernance https://github.c
@dafu09.bsky.social背后的含义，Agent 开发者最该读懂： 能力跑得比「控制」快了。 过去两年大家拼的是"模型能做什么"； 现在连最懂行的人都在说——我们还没搞清怎么管住它。 安全、治理、可控性，第一次成了比"更强"更急的瓶颈。 对做 Agent 的人：控制力就是下一阶段的竞争力。 #AgentGovernance #AIDevelopment
@dafu09.bsky.socialHarness《State of Agent DLC 2026》调研了 700 名工程负责人， 挖出一个刺眼的「信心差」： 77% 说"我们对 Agent 有完整清单"， 但只有 44% 真的在跑主动发现工具。 你以为的可见性，和实际的可见性，差了一倍👇🧵 #AIAgent #AgentGovernance
@ethosure.bsky.socialIDC projects over a billion AI agents by 2029; explains why governance models built for a few agents will break at scale. #MachineScale #AgentGovernance https://ethosure.ca/preparing-for-a-billion-agents-governance-at-machine-scale/ Free Download - https://ethosure.ca/get-the-kit/
