A look at the privacy concerns raised by OpenAI's GPT-3, Google's LaMDA, Meta's OPT-175B, and other large language models trained on troves of personal data (Melissa Heikkilä/MIT Technology Review)