Harm Over Failure
- Author
- Justin Mayer & Dan Jacobson
- Published
- Thu 24 Jul 2025
- Episode Link
- https://shows.arrowloop.com/@abstractions/episodes/019-harm-over-failure
Large language models will commit blackmail and murder to achieve goals, LLMs know a lot about us, and some managers mandate employees use LLM tools to do their jobs.
- Historic web nostalgia
- Agentic misalignment
- What LLMs know about their users
