OpenAI points out that a new variety of self-replicating prompt injection is possible.
- OpenAI points out that a new variety of self-replicating prompt injection is possible.
- Hold onto your butts!
From other episodes
- Tweet: "Had my first semi-uncomfortable Instinct interaction.2026-10-12
- A tweet about Instinct's absurdly bad handling of prompt injection:2026-09-28
- This week in the Wild West Roundup:2026-09-28
- It seems most likely that OpenAI didn't look at the chat logs of the other researcher to solve Navier Stokes.2026-09-21
- This week's Wild West roundup:2026-09-21
- Remember, with prompt injection, even an email you don't read could infect your agent and make it a zombie.2026-09-21
- This week in the Wild West roundup:2026-09-15
- The danger of prompt injection grows super linearly as a thread gets longer.2026-09-15