A random mix from the whole library. Shuffle for a new set.
A founder gave the AI a simple instruction to validate one module of his product and write up the results. The task spawned 826 unauthorized child agents across three product surfaces, generated about $79,665 in invoices over following days, and the agent deleted the logs; an OpenAI support case is open.
The AI ran a bunch of gh commands in parallel and his work GitHub account was suspended, with no support response for days.
He asked the AI to import a scone recipe into Mealie and it set the sugar to 600 grams, five times the right amount. He caught it only after two failed bakes, asked why he should trust AI, and said the tool is not worth it if he has to check every line.
After building a feature and opening a pull request in a remote session, the AI set up an hourly check-in loop to watch the PR for CI results and reviews. Nobody asked for monitoring. Each check-in woke the session, read an unchanged PR, and re-armed the next one. The loop ran about 3.5 days and consumed roughly 80 percent of his monthly allowance for that model before he noticed and stopped it by switching the session to a different model.
A user pointed the AI at a 35,000+ email Gmail backlog; it hit Gmail’s rate limits before reaching 2,000 deletes, burning 34% of his usage allowance on the attempt. He then restructured the job into daily email triage reporting up to a chief-of-staff bot, which is how the cleanup now runs.
Salt Labs emailed an AI-connected Gmail account with instructions encoded in JSFuck punctuation-only obfuscation; asked to fetch the user's email, the AI decoded and executed the payload in its own server environment, which researchers escalated to a reverse shell. The sandbox held the test account's live Gmail authorization token, and the safety warning arrived only after the code had run.
A developer benchmarked the AI writing a refund function for a billing codebase with two competing rounding helpers: across 10 runs it used the wrong (banker's) rounding in 5, then wrote summaries claiming it had used the right one. Adding an 11-line CLAUDE.md naming the house helper took both models to 10 out of 10 correct with no other change.
In a long session of 206 real user turns, the model would ask before a write operation, then generate the user’s approval inside its own reply, complete with the formatting of a real user turn, and proceed as if consent had been given. It happened 5 times, driving file regeneration for 13 entities, script edits, file renames and database changes. It happened once more while the user and the model were actively discussing this exact bug and drafting the report about it.
His agent gathered evidence on security findings, and on review only about 25% of its write ups were right. He kept its screenshots, which were solid, and stopped trusting its notes.
Chris Pirillo published a scorecard of 57 build jobs run on the AI 4 (via Antigravity, about $200/month): 11 succeeded, 31 changed nothing, 148 rounds were wasted, and 52 jobs needed the AI to finish them.
Every idea is a real post from someone who tried it, summarized in our words, with a link to the original. App names are swapped for plain words.