TLDR Tech / AIWhat happened after 2,000 people tried to hack my AI assistant2026-06-29Prompt injections are still a real security problem, but defenses are getting better.Read the full story