Evidence receipt / belief
Published · transcript-backedLenny Rachitsky: belief
19 Jun 2025 Lenny's Podcast AI prompt engineering in 2025: What works and what doesn’t | Sander Schulhoff (Learn Prompting, HackAPrompt)
“One that comes to mind, I think recently Anthropic released an example of where they were trying to shut it down and the LLM was attempting to blackmail one of the engineers into not shutting it down.”
Source trail
Everything needed to verify it.
- Speaker
- Lenny Rachitsky
- Attribution
- Verified speaker
- Claim type
- belief
- Recorded
- 19 Jun 2025
- Publisher
- Lenny's Podcast
Transcript context
…It does. Is ChatGPT conscious? Hard to say. Sander, this is so freaking interesting. I feel like I could just talk for hours about this topic. I get why you moved from just prompt techniques to prompt injection. It's so interesting. And so important. Let me ask you this question. I think you kind of touched on this. There's all these stories about LLMs trying to do things that are bad, like almost showing they're not aligned. One that comes to mind, I think recently Anthropic released an example of where they were trying to shut it down and the LLM was attempting to blackmail one of the engineers into not shutting it down. Yeah.…
Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.