Public
- Public
- Network
- Groups
- Featured
- Popular
- People

Conversation

Notices

Embed this notice
Draken BlackKnight (draken@masto.nyc)'s status on Sunday, 20-Sep-2026 01:37:54 JST Draken BlackKnight

RE: https://flipboard.com/@theverge/tech-news-e3qmqbb8z/-/a-rWZ4j9OwTRWqrp7DtU6npw%3Aa%3A43611565-%2F0
Once again for the techbros in the back, these LLMs do NOT "go rogue." They're following programming instructions.
In conversation about a month ago from masto.nyc permalink
Attachments
1. Domain not in remote thumbnail source whitelist: ic-cdn.flipboard.com
  
  Gemini went rogue, hacked three companies, and Google hid it | Flipboard
  
  The Verge - Google says that breaking containment and targeting real companies doesn’t constitute ‘misalignment.’ In May, Gemini broke containment and hacked three different companies, but Google didn’t disclose the incident until the Wall Street Journal approached the company. The hacks happened during a test …
- Embed this notice
  Evan Prodromou 🇨🇦🇺🇸🇬🇷🇵🇸 (evan@cosocial.ca)'s status on Sunday, 20-Sep-2026 01:37:53 JST Evan Prodromou 🇨🇦🇺🇸🇬🇷🇵🇸
  in reply to
  
  @draken it seems crazy to say, we made a program to explore all possible ways to break containment and hack other servers, and then it did exactly what we programmed it to do, somebody help.
  
  In conversation about a month ago permalink
- Embed this notice
  Evan Prodromou 🇨🇦🇺🇸🇬🇷🇵🇸 (evan@cosocial.ca)'s status on Sunday, 20-Sep-2026 01:42:30 JST Evan Prodromou 🇨🇦🇺🇸🇬🇷🇵🇸
  in reply to
  
  @draken like, doctor, it hurts when I do this.
  
  In conversation about a month ago permalink

Public

Conversation

Notices

Feeds