All stories
41 publishedGoogle's chatbot got an astronomy fact wrong in its own launch advert
The error was in the promotional material, spotted by astronomers before the launch event had finished.
Six court cases, correctly formatted, complete with judges. None of them existed.
A New York lawyer filed a brief citing six precedents ChatGPT had invented — with plausible names, docket numbers and quoted opinions. He then asked the same chatbot whether the cases were real. It said yes.
A model trained on 2,500 molecules found an antibiotic that screening had walked past
It had been sitting in a library of known compounds, investigated as a diabetes drug and never pursued as an antibiotic.
Seventeen doctors over three years. The answer came from a chatbot — and a surgeon confirmed it.
A mother typed three years of her son's symptoms and MRI notes into ChatGPT. It suggested tethered cord syndrome. A paediatric neurosurgeon looked at the scans and agreed.
Air Canada argued it was not responsible for its own chatbot. The tribunal disagreed.
Air Canada's chatbot told a grieving passenger to book at full price and claim the bereavement rate within ninety days. The airline's real policy excluded claims once travel was over. A tribunal ruled it was answerable for what its own chatbot said.