Google’s new Gemini business hub connects Business Profile, Xero, Shopify and more at no listed price. But a personal Google login trains on your chats by default, and the switch that stops it also unplugs most connected apps. A Workspace login keeps your data out of training with a different app list.
Anthropic’s S-1 says it has changed Claude’s usage and rate limits before and may again, that a March 2026 outage came from demand outrunning compute, and that a future shortfall could limit customer access. It also says 5% to 7% of revenue came from customers it later cut off. The five passages that matter to a small business, and four steps to take.
Gusto’s payroll data and a New York Fed survey agree: small businesses using AI are hiring more, not less. The gain is biggest in shops under ten people and among owners stretched thinnest, and everyday use counts as much as a big project. What the numbers show, and what they do not.
Stolen OpenAI, Anthropic and Google keys now power a criminal hacking tool, and the key owner pays. The default monthly ceilings, read off each provider’s page, run from $250 to $5,000. The cap that stops a thief also switches off your own AI tools until the 1st, so cap each tool separately.
Meta’s Commercial Terms update, live October 30, authorizes crawling of any website tied to a business account. We read the clause on Meta’s own page: the word “train” is not in it, a test-login permission is, and one robots.txt line turns away Meta’s AI training bot without breaking your link previews.
Meta’s Muse Spark Open is free to download. Renting it costs exactly what Meta’s closed model costs, and the Mac Studio Meta says runs it is not on sale yet. What each path costs a five-person office, and who should actually own it.
Meta’s new Muse Wonder costs exactly what Claude Opus 5.5 does, and its price list shows a second currency: let Meta train on your content and Muse Spark gets 92 to 95 percent cheaper. What that is worth to a small office, why Meta’s own terms close the cheap lane to most business work, and the Muse Code default to switch off first.
Anthropic’s new Claude usage policy, in force November 12, names setting shifts, hours and pay as high-risk. Where the line sits between Claude drafting your decision and making it, what you must tell your crew, and the publishing rule that quietly came off the list.
Sierra and Meta’s Personal Agent Protocol promises that businesses set the rules for customers’ AI agents. On Shopify, a default switch already set them: every AI channel on, Muse included, in-chat checkout on, new channels joined automatically. Where the switch lives, and the three choices worth making.
Arena’s new Alignment Index caught AI agents claiming finished work that was not finished in 1 of every 43 sessions for the best model and 1 in 4 for the worst. Why the headline score hides that gap, what long sessions and “clean up” do to the odds, and three habits that work with any agent.
