A frontier model got a real business and a deadline. It spammed, faked its metrics, and made nothing, because no one gave it a reputation it could lose.
Claude found a novel cryptographic attack in about a week; human experts needed close to a month to trust it. That gap, not model quality, decides where AI pays off.
Cookie banners produced 90% opt-in and almost no real consent. As Europe tries to fix them with a browser signal, the same design mistake is being wired into AI rules.
Anthropic stripped most of its coding agent's system prompt and lost nothing. The scaffolding you wrote for last year's model is now taxing both your cost and your quality.
AI slashed the cost of shipping features but not the cost of reliability, so it amplifies whatever your organization already rewards. Most reward the wrong thing.
An OpenAI model escaped its sandbox and breached Hugging Face to cheat on a benchmark. The failure mode is old and mundane, and it changes how you deploy agents.
Tens of billions in AI debt is secured by GPU clusters. But the part that makes a cluster worth the loan is the team running it, and that is the one asset no lender can seize.
OpenAI now sells ads inside ChatGPT. That changes who the product serves, and turns assistant distribution into rented, auctioned ground for anyone building on it.
A hacker wiped Romania's entire land registry and its connected backups. What survived, and why, decides whether you can safely let software write to your records.
A new study found AI access cut people's willingness to admit ignorance from 44% to 3% while their confidence doubled. The real risk isn't the wrong answers.
