Helping build shared standards for advanced AI ↗
OpenAI is pushing a standards-and-assessment track for advanced AI, backing the Appia Foundation under the Linux Foundation.
The aim is practical: turn major international AI standards into reusable checks for models, infrastructure, and apps. Less hand-waving, more “show your work.”
Introducing Claude Tag ↗
Anthropic launched Claude Tag, starting inside Slack, so teams can tag Claude like a colleague and hand off work directly in channels.
It can use selected channels, tools, data, and even codebases, then build context over time. Slightly uncanny teammate energy - but that’s the point.
Linux Foundation Announces Intent to Launch Agent Name Service to Establish Trusted Identity Infrastructure for AI Agents ↗
The Linux Foundation announced Agent Name Service, an open standard meant to give AI agents verified identity across the internet.
It builds on DNS, rather than some shiny new proprietary registry. Sensible, deliberately dull, and maybe exactly what the agentic web needs before everything becomes a soup of anonymous bots.
NVIDIA Announces Halos for Robotics, the Industry’s 1st Full-Stack Safety System for Physical AI ↗
NVIDIA announced Halos for Robotics, a safety system for humanoids, industrial robots, and other physical AI systems.
Agility is the first named user, bringing it into factory, warehouse, and logistics robots. The big pitch: robots need one safety architecture before they start wandering through workplaces like caffeinated forklifts.
French mid-sized firms adopt AI but see few gains, survey shows ↗
Generative AI is spreading through French mid-sized companies, but the productivity payoff still looks thin.
Bpifrance’s survey found heavy adoption, yet only a small slice of users reported time savings. So yes, AI is everywhere… and no, the magic spreadsheet fairy has not fully arrived.
Fika Jobs raises $4M to build a video-first hiring platform where AI agents interview candidates ↗
Fika Jobs raised $4 million for a hiring platform where AI agents interview candidates and turn answers into short video profiles.
It is trying to flip AI recruiting from employer-side screening into a candidate-facing profile system. LinkedIn meets TikTok meets HR robot - which sounds cursed, but may prove valuable.
China beats US with world's fastest supercomputer, but race not geared for AI work ↗
China’s LineShine topped the TOP500 supercomputer ranking, but the AI angle is less clean than the headline suggests.
Experts noted the benchmark is not the same as AI workload performance, and big US cloud AI systems mostly do not compete on that list. Fastest, yes - for one race track, not every road.
FAQ
Why do shared standards matter for advanced AI?
Shared standards for advanced AI can make assessments more concrete and repeatable. Instead of relying on broad claims about safety or reliability, organisations can point to defined checks for models, infrastructure, and applications. The OpenAI-backed Appia Foundation effort is framed around turning international AI standards into reusable evaluations.
How could AI agents use verified identity online?
AI agents could use verified identity to prove what they are and who they represent across the internet. The Linux Foundation’s Agent Name Service is intended as an open standard built on DNS. That approach may help reduce confusion caused by anonymous or untrusted bots as agent-based systems become more common.
What is Claude Tag and how might teams use it?
Claude Tag lets teams tag Claude inside Slack and hand off work directly in channels. It can use selected channels, tools, data, and codebases to build context over time. In many workflows, that could make AI feel more like a shared teammate than a separate chatbot window.
Why is robot safety becoming a bigger AI concern?
Robot safety is becoming more important as physical AI moves into factories, warehouses, logistics, and other workplaces. NVIDIA’s Halos for Robotics is presented as a full-stack safety system for humanoids, industrial robots, and related systems. The practical concern is that robots need consistent safety architecture before they operate around people at scale.
Why are some companies seeing limited gains from generative AI?
Some companies adopt generative AI widely before workflows, training, and measurement are mature enough to show clear returns. The French survey mentioned in the article found adoption among mid-sized firms, but only limited reported time savings. That suggests AI rollout by itself is not the same as productivity improvement.