In partnership with
Welcome back!
This week had everything: an AI art protest that turned into an arrest, a viral open source “Jarvis” people are running at home, and Claude quietly slipping into Excel.
Also, the money question is getting louder. The models are impressive. The bills are, too.
⚡ Quick Overview
He ate the AI art and got arrested: a campus show turns into the most literal AI protest yet.
A viral open-source Jarvis hit 10K stars: Clawdbot runs 24/7 and you can text it tasks.
Claude lands in Excel for Pro users: the spreadsheet gets a copilot with guardrails.
OpenAI bankruptcy warnings: growth is real, but so is the burn rate.
Want to get the most out of ChatGPT?
ChatGPT is a superpower if you know how to use it correctly.
Discover how HubSpot's guide to AI can elevate both your productivity and creativity to get more things done.
Learn to automate tasks, enhance decision-making, and foster innovation with the power of AI.
A STUDENT ATE AI ART AND GOT ARRESTED
What’s Happening
At the University of Alaska Fairbanks, a student allegedly pulled AI-generated images off a gallery wall and chewed them in protest. Reports say 57 of 160 pieces were damaged, with the student chewing and spitting them out. He was arrested and charged with fifth-degree criminal mischief (a misdemeanor tied to low-dollar damage).
The exhibition itself was created by an MFA student and framed as a project about identity and “false memories,” with the artist saying it was connected to his experience with “AI psychosis.” The artist later dropped personal charges, but the case still moved forward through the state.
Why It Matters
This is the most literal version of the argument everyone has been having online.
This is the culture clash in physical form. A policy argument turned into property damage and an arrest.
The protest was the message. The student said it was meant to reflect how AI “consumes” other people’s work.
Expect more real-world blowups. As AI content becomes normal, the social rules around credit and consent are still being negotiated.
“Performance art” is not a legal defense when someone else’s work gets destroyed.
What’s Happening
A self-hosted, open source assistant called Clawdbot has been going viral, blowing past 10,000 GitHub stars in days. The pitch is simple: run it 24/7, connect it to your tools, and text it instructions like you would a human assistant.
Early users say they’re using it for morning briefings pulled from email and calendar, reminders, follow-ups, and basic web tasks. It’s also sparked a mini trend of people buying Mac Minis specifically to run it locally as a low-power always-on agent.
It’s new, rough around the edges, and not a polished consumer product, which is part of why the warnings matter: it needs deep access to do real work, and that comes with real risk.
Why It Matters
This is one of the first “agent” stories that feels like people are actually living with it.
Always-on is the unlock. The value is not a smarter chat. It’s something that sticks around and handles little tasks all day.
Open source makes it spread fast. People can tinker, remix, and run it their own way, which is how tools become movements.
It also comes with sharp edges. If an agent touches your inbox or browser, a small mistake can turn into a big one.
Rule of thumb: treat it like giving a new intern access to your accounts.
CLAUDE IN EXCEL EXPANDS TO PRO USERS
What’s Happening
Anthropic expanded its Claude for Excel integration to Claude Pro users after it previously being more limited. The idea is not just chatting about your spreadsheet. Claude can actually work inside it, reading workbooks, making edits, and generating new sheets.
Recent updates highlighted in coverage include things like dragging in multiple spreadsheets, safeguards to prevent overwriting existing cells, and longer sessions that don’t fall apart when the workbook gets big. Some versions also emphasize transparency, showing what changed and why, so you can verify the work instead of trusting it blindly.
Why It Matters
Spreadsheets are where decisions get made, and where mistakes get expensive.
This hits the most common “real work” format on earth. Budgets, forecasts, headcount, planning, and ops all live here.
Speed is great, but review is the real feature. The best AI in Excel is the one that leaves a clean trail you can audit.
It changes who can do what. More people can build models, clean data, and test scenarios without being an Excel wizard.
If AI gets good at spreadsheets, it gets good at the adult world.
FINANCIAL EXPERTS WARN OPENAI MAY GO BANKRUPT BY 2027
What’s Happening
A new wave of commentary is questioning whether OpenAI’s spending curve is sustainable long-term, even with massive funding and huge usage. Analysts and financial observers have pointed to the same tension: frontier AI demands relentless compute, and the cost of scaling can rise faster than revenue.
Some warnings frame it dramatically, even using “bankruptcy” timelines as a worst-case scenario if spending stays aggressive and new monetization does not keep up. OpenAI isn’t bankrupt, but the fact that this conversation is happening at all says a lot about how expensive the frontier has become.
Why It Matters
This is bigger than one company.
If the biggest AI product struggles to pay the bill, everyone else feels it.
It explains why monetization is changing. Subscriptions, enterprise deals, usage limits, ads, all of it.
It decides what AI becomes for normal people. If it’s costly, access tightens. If it gets cheaper, it becomes everyday infrastructure.
The next advantage is boring but decisive: lower cost per answer.
THE BIGGER PICTURE
AI isn’t just getting smarter. It’s getting embedded.
It’s showing up in classrooms, inside inboxes, in spreadsheets people run companies on, and in budgets that don’t care about hype.
The next phase is less about what AI can do and more about what we’re willing to trust, pay for, and hand access to.
If this issue helped you make sense of AI’s chaos, forward it to a friend who shouldn’t be sleeping on this.
What did you think of today's edition?
Until next time,
Long Live AI