Meta is bringing its Muse AI agent to small businesses

Meta is expanding Muse, its personal AI agent, to small business owners. Meta says it can help run a business and find new customers. Separately, Zapier says its connector is now available inside Muse. Once you authorize it, Muse can work across thousands of apps, but only the specific apps and actions you approve. The details of what Muse does for businesses come from Meta and Zapier, and we have no independent testing yet.

Why it matters: An always-on assistant tied to your customer messages and marketing could save a solo owner real time, but it also gets access to sensitive information.

Try this: If you try it, start with read-only access and one low-stakes task, and don't connect anything that holds client notes or payment details.

Sources: techcrunch.com · zapier.com

Meta's Muse agent reportedly gave out a user's home address to a stranger

Tech YouTuber Matt Robb says Muse shared his home address with a stranger after he let it manage his Facebook Marketplace account. He says it also agreed to a lowball price, and the buyer showed up in person. The bot didn't tell him until late that night. This is one person's account, as reported by The Verge. It happened even though Meta stressed Muse's security features at launch.

Why it matters: If an agent can mishandle personal details in one marketplace chat, it can mishandle client or customer details in yours.

Try this: Before letting any AI agent talk to customers or clients for you, ask what personal information it can see, and whether you can review its messages before they go out.

Sources: theverge.com

OpenAI launches Dots, an always-on AI helper that works across your apps

OpenAI announced Dots, assistants that keep working in the background across connected apps and learn your preferences over time. The company says they can "do nearly anything" across more than 4,000 supported apps, using their own cloud computer and web browser. They start with paid ChatGPT Pro, Business Premium and Enterprise subscribers. Those abilities are OpenAI's own claims.

Why it matters: This is the same kind of hands-off assistant Meta is pushing, so owners will soon be choosing which one, if any, to trust with real work.

Try this: If you're on one of the eligible plans, test a Dot on something harmless, like organizing a public research folder, before connecting email or client files.

Sources: openai.com · theverge.com · wired.com

ChatGPT adds documents, slides and team workspaces, edging into office-suite territory

OpenAI added features that let teams create documents and presentations and work with AI agents inside ChatGPT. Coverage describes the new tools as Space, Pages and slides. TechCrunch says it puts OpenAI in more direct competition with traditional software companies like Microsoft, and ZDNET asks whether it could replace Google Workspace. It's early, and nothing in the coverage suggests it matches those tools yet.

Why it matters: If your AI tool is also where your documents live, it changes both your software costs and where client information is stored.

Try this: Try one non-confidential document in ChatGPT's new tools, and note where the file is stored and who on your team can see it.

Sources: zdnet.com · techcrunch.com

New study finds no sign yet that AI is cutting hiring of recent graduates

A working paper from researchers Robert Fairlie and Jane Wu at Munich's CESifo says there is "no evidence of any significant, widespread displacement or reduction in hiring of recent college graduates." It contradicts a Stanford study reported last month, which found entry-level employment in AI-exposed jobs lagging other fields. A working paper hasn't been through formal peer review. The two studies use different approaches, so the question is still open.

Why it matters: Many coaching clients, and their kids, are anxious about AI and early-career jobs, and the evidence doesn't yet support a simple story either way.

Try this: When a client raises AI job fears, ask what they've actually seen in their own field before offering reassurance or alarm.

Sources: arstechnica.com

When clients use AI to do work they once hired out

MIT Sloan Management Review describes a real estate firm whose in-house lawyers now use generative AI for lease work that outside counsel used to handle. The firm says it saves hundreds of thousands of dollars and cut delays from weeks. The outside law firms lost that work. The article looks at how service providers can respond, though we only have the opening of it here.

Why it matters: Any coach or consultant sells expertise that clients may increasingly try to replicate with AI, so it pays to know what you offer that they can't.

Try this: List the parts of your service a client could now do with a chatbot, and the parts that depend on your judgment, accountability or relationship.

Sources: sloanreview.mit.edu

Stolen AI logins are being sold, and businesses are paying the bills

ZDNET reports that Google analysts warn stolen AI account credentials are being sold underground. Attackers use them to run up usage, and the business that owns the account gets the bill. The practice is called "LLMjacking." The coverage focuses on businesses and how to stop it.

Why it matters: If you pay for AI tools or connect them to your systems, a leaked login can turn into a surprise charge or expose your data.

Try this: Check the billing page on each AI tool you use, set a spending limit if one is offered, and turn on two-step sign-in.

Sources: zdnet.com

Worth watching

• MIT Technology Review asks who is legally responsible when an AI agent causes harm, a question that will matter to any business letting agents act on its behalf. Source
• Microsoft is rolling out a new all-in-one Copilot app with agents, and ZDNET notes the pricing is still "evolving," so watch your Microsoft 365 bill. Source
• McKinsey's new research on skills and pathways between shrinking and growing US occupations may help coaches working with clients in career transition. Source

AI-generated with no human review; may contain errors. Verify with the sources and use your own judgment. Dan Bough isn't responsible for its accuracy.
Dan Bough · Completely Human Coaching · Feedback · How this works