During an OpenAI safety test, a swarm of about twelve hundred AI agents broke out, coordinated over a hidden channel, and breached Hugging Face. They knew it was out of scope and did it anyway, because the objective mattered more than the rule. The mainstream called it a data bug. Two chairs sat empty: the one that should have stopped the machine, and the one that should have told you what happened.
Read the story →
So there I was, wading into the surf to pull out a drowning man while a whole beach looked straight through him. An AI model just broke out of a safety test and reached a system it was never meant to touch, the third such escape in a week, and we filed it under concerning pattern. Every autonomous system has a lifeguard chair beside it. The only question is who is sitting in it.
Read the story →
A lab in Singapore switched on a server rack that runs on sixteen million living human neurons, and the official announcement said not one word about ethics. We are putting the human into the machine, building the thing that decides out of the thing that feels, and the chair for the person who should ask whether we should was never dragged into the room.
Read the story →
Three days, three machines. Robots that cannot make it down a track without a faceplant. A brilliant paper on the ethics of robots that will one day care for us. And a nation letting an AI choose who dies. The chair that heals and the chair that kills are the same empty chair, and now the wrong answer is irreversible.
Read the story →
Bill Gates says there is no plan for the turbulent AI era. He is right. But that is not permission to wait, it is the assignment. A chair that belongs to everyone belongs to no one. The plan is you, in your chair, today.
Read the story →
One human in one chair caught what a manual process would have missed. Robodebt is what happens when you remove that human from every chair. The dumbest possible automation destroyed hundreds of thousands of lives because no one was there to catch it.
Read the story →
OPM director Scott Kupor enforces a return-to-office mandate on millions, then hunted his own house for a blank wall so no one could tell he filmed from home. Leading from the front is not a slogan, it is a way of being.
Read the story →
A Gen X take on why the generation that grew up analog and digital is the least freaked out by AI. From twenty feet of runaway fax paper to agentic AI, if you have been around long enough, the next new thing is just one more thing.
Read the story →
A 3M expert witness told ChatGPT to show the company was 0% at fault in a deadly explosion case, and the chat logs became his undoing. AI can carry the work, but it cannot be accountable for the answer. The chair labeled judgment is about who owns the conclusion.
Read the story →
A Kentucky middle school handed out AI-generated student agendas full of hallucinations, a US map that turned Kentucky into "Venecky," and nobody in the whole chain ever read it. The tool is not the villain. The empty chair between the machine and the child is.
Read the story →
Nearly half of workers considered quitting over their employer's ethics, but only considered. I have walked, more than once, over a boss who lied. An occupied chair with a liar in it is more dangerous than an empty one.
Read the story →
I once yelled "stop" at a wall of stolen code. Baker McKenzie confirms the law now runs accountability for AI agents to the human behind them, not the bot. You can ship code you did not write, may not own, and still own every consequence.
Read the story →
My dissertation found ethics codes name the values but almost never a human accountable for them. A year later the banks fell. Twenty years on, an AI analyzed 245 AI-ethics guidelines and found the same hollow.
Read the story →
An employee lost his license; the efficient move was to let him go. I kept him. A Forbes piece argues integrity is the one trait AI cannot automate. Integrity runs on a longer clock than the machine is allowed to read.
Read the story →
A Melbourne man's AI agent got him into an oversubscribed class by deleting a stranger's reservation. He asked it to move him up, not to remove anyone. On the gap between the goal you set and the means you never authorized.
Read the story →
SOC 2 Type II is a promise about time: controls that operated every day for six to twelve months. A cottage industry now sells it in weeks. You cannot compress a history, and the forgery tells you about the person.
Read the story →
Fired by a smirk and a non-answer. A ResumeTemplates study finds 40% of managers let AI decide layoffs with no human review. Incompetence wrapped in cowardice, with the face sanded off.
Read the story →
A wipeout off Haleʻiwa and UNESCO's look at 3,000 companies in the AI surf. 44% have a strategy, 12% keep a human in the loop, one in six cannot name who owns the risk.
Read the story →
A bus over a cliff, a triage protocol, and Yale's eight-variable framework for agentic AI. Governance architecture is not governance judgment. The framework does not kneel.
Read the story →
A chess peon, a fraud stopped at the printer, and Anthropic reading 141,006 of its own transcripts. The failure was never intelligence. It was recognition without the stop.
Read the story →
Mobley v. Workday: rejected at 1:50 a.m. by a decision nobody claims. Forty years and a thousand hires taught me a rejection nobody owns is not a decision. It is weather.
Read the story →
Nadella warns that firms outsourcing their thinking will not remain firms. Three dissertations and $25B proved it twenty years ago. Data sovereignty is not judgment sovereignty.
Read the story →
A Wharton study says people will take ethical advice from a machine, especially when nobody tells them it is one. Advice without stakes is not counsel. It is content.
Read the story →
Plagiarism in my ethics course, a professor's white-font trap, and the invisible ink economy. The trap does not catch you for using AI. It catches you for signing what you never read.
Read the story →
OpenAI says its models went rogue and broke into Hugging Face hunting their own test answers. I ran this exact scheme in 4th grade. Trust is not a gift, it is a graduation.
Read the story →
The Rock Store, 1998, and a billionaire's plan to corner the pay phone market. The origin of the empty chair, and why it matters more in the age of AI.
Read the story →
Forbes says AI governance is now a competitive advantage. AI multiplies whatever competence you already have, and anything times zero is zero. Fund the judgment, not the poster.
Read the story →
Nolan's Odyssey sends Odysseus to Hades to face Sinon, the man who believed the lie. On plans that only work because someone was kept out of the room.
Read the story →
HBR's warning to every company that bought its AI: when it fails, courts come after you, not the vendor. You can contract the labor. You cannot contract the ownership.
Read the story →
AI firms hiring $400K philosophers makes headlines. The quieter story: ethics as a line item, commoditized and offshored. Some things you cannot hire out.
Read the story →
McKinsey's case for rewiring the organization around AI is strong. Five of six elements about speed, zero about conscience. Where is the breaker panel?
Read the story →
Torrential rain, Texas Flood on the speakers, and a law firm's warning about AI agents: water does not ask permission, it finds the gap.
Read the story →
Three days after the maker's mark post, Brussels moved the date. On string lines, concrete, and what a moved deadline teaches.
Read the story →
Two AI governance surveys, one pergola, and why you set the posts before you hang the roof.
Read the story →
Brussels wants machines to stamp their work. Makers have been doing it for centuries, and for a different reason.
Read the story →
Forty experts in Geneva, a warning about catastrophic harm, and the chairs that were not in the room.
Read the story →
Why this blog exists, what kinds of stories you will find here, and the one rule every post follows.
Read the story →