The Morning PaperSaturday, September 26
AI models can't tell when a hacker put words in their mouth
AI models can't reliably tell when hackers inject fake instructions into their prompts—and training them to recognize it often backfires.
~70s readRead today’s paper. Start a streak.
