Spotify Lets You Edit Its AI Profile of You
September 23, 2026The through-line today isn't any single story — it's control. Who has it over AI systems, who's losing it, and who's quietly handing it away.
Spotify Finally Shows You What It Thinks You Are
Spotify is rolling out something called Taste Profile to U.S. Premium subscribers: a window into how its recommendation algorithm has been reading your listening habits, and a way to correct it using plain language. If the algorithm decided you're a "sad indie folk" person because you played one Phoebe Bridgers album in 2023, you can now tell it otherwise. This is what "explainable AI" looks like when it's built for regular people — systems that show their reasoning in human terms, not just produce outputs — and it's a design choice, not a technical achievement. That choice is going to age other platforms poorly.
Meta Shipped an AI Assistant With a Hole in the Wall
Meta's new Muse AI assistant launched on Mac with what security researchers call a zero-day vulnerability — meaning a flaw that existed from day one, before anyone had a chance to patch it. Specifically, an attacker who exploited it could execute arbitrary code on the victim's machine: open files, exfiltrate data, run software, all without the user knowing. Meta says it's fixed. But the episode is a useful reminder that AI assistants aren't just chatbots anymore — they're agents (software that can take actions on your behalf, like opening files, sending messages, browsing the web), and agents with bugs aren't just annoying, they're entry points for attackers into your whole system. If your company is evaluating AI tools that can act on your systems, the question to ask vendors isn't just "what can it do?" It's "what can someone else make it do?"
https://www.wired.com/story/metas-muse-ai-agent-zero-day/
AT&T Is Using AI as a Reason to Cut People and Sell It as Efficiency
AT&T is telling Wall Street it can operate with significantly fewer employees by automating large portions of its network operations and customer service. The framing — fewer people, less electricity, same or better output — is going to become a standard corporate script. What makes this story worth reading is how candid AT&T is about the sequence: automate first, then justify. That's the opposite of the "AI augments workers" story most companies tell publicly. In AT&T's investor materials, the logic isn't that AI makes employees more productive — it's that AI makes certain employees unnecessary. The gap between what companies say about AI and workforce and what they're committing to on earnings calls has never been wider, and it's worth knowing which version your own leadership is telling internally.
https://www.wired.com/story/atandt-is-automating-away-its-old-telecom-empire/
OpenAI Is Coming for the Creator Economy
OpenAI has hired three senior executives from Patreon — including co-founder Sam Yam — to lead something the company is calling Creator Product. Patreon built the infrastructure for independent writers, podcasters, and artists to earn directly from their audiences. The fact that OpenAI is pulling talent specifically from that world suggests it's building something closer to a creator monetization platform than another chatbot feature. If you sell knowledge or content for a living, OpenAI is not just a tool vendor. It is positioning itself as a platform that sits between creators and their audiences, and that's a different kind of relationship entirely.
A Prime Minister Admitted Governments Are Already Behind
Greece's PM Kyriakos Mitsotakis said something most leaders won't say on the record: governments are "fighting yesterday's battle" on AI regulation, and no government is actually prepared for what's coming. He didn't offer a solution. That honesty is itself the story. Regulatory frameworks being developed right now are largely designed around AI as it existed two or three years ago — image generators, early chatbots. The AI being deployed in 2026 can take actions, negotiate, automate entire job functions, and integrate into infrastructure. The mismatch between the rules being written and the technology being shipped is significant, and the people most exposed to that gap are the professionals and businesses making decisions in between.
The control you don't think to ask for is the control you've already given up.
Stop staring at the blank ChatGPT box
The Plain-English AI Prompt Pack: 100 copy-paste prompts for real work — emails, meetings, marketing, managing people. Works with any chatbot.
Get the Prompt Pack →Get the daily brief
Plain-English AI news for people with real jobs. Free, three minutes a day.