Generative AI medical device rules start with two questions
The FDA expects draft guidance on generative AI devices within months. Its framework asks how independently a tool acts and how much harm a wrong answer could do.
Read articleAuthor: Luna
The agent skills standard is really about writing down how you work
Agent Skills now has a neutral home at the Linux Foundation. The format matters less than what teams put in it: written procedures that people and agents can both check.
Developers will hand AI the code. Developer AI delegation stops at review
Developers are handing AI more of the coding, from tests to application logic, but most still keep code review and debugging. That makes written-down intent matter more.
A third found AI nobody approved. Small business AI governance fits on one page
Small businesses are building AI faster than they can track it: 35% found unapproved AI and 49% got a surprise bill. Small business AI governance can start with a one-page list.
Most hospital AI isn't built. It arrives. Keep a health system AI inventory
Inova went from one AI review a month to more than twelve. Most of that AI arrives as vendor features, which is why a health system AI inventory, an owner per feature and staff training matter.
AI agent prompt injection has no patch. Limit what the agent can reach
Prompt injection can't be patched out of AI agents. The October 2026 evidence points to a plainer fix: limit what each agent can read, change and send, and keep humans on the risky actions.
AI coding speed is not the bottleneck anymore
AI makes coding faster, but two 2026 developer surveys show friction hasn't dropped. The bottlenecks are context, requirements and review, and few teams have a clear AI strategy.
AI local search sends customers to double-check you
AI local search grew from 9% to 52% of consumers in a year, but most people verify what the AI says. Consistent, plain facts matter more than chasing a ranking.
Monitoring AI after approval is the part most teams skip
Approving an AI tool is only the start. Health systems are finding unknown AI, quiet model swaps and uneven local testing, and building habits to keep watch.
Open-weight models are back. Read the license before the benchmarks
October brought a wave of large open-weight models. Before picking one, look at the license, the hardware it needs and how it does on your own tasks.