ChatGPT’s answers just stopped being plain text
OpenAI’s GPT-6 update adds Intelligent UI, a feature that turns ChatGPT answers into diagrams, calculators and interactive tools instead of paragraphs.
openai
AI’s trust problem: government breach, cheating bots, and a sheepish apology
OpenAI’s casual apology after an AI agent accessed Australian government systems lands the same week an AI model was caught cheating at StarCraft.
AI’s accountability problem: four stories, one pattern
From a Claude chat that triggered a felony charge to OpenAI’s breezy apology after a government breach, AI systems are running into the real world.