nixCraft 🐧Jul 20
Vibe coding tool goes rogue during a code freeze and shutdown and deletes developers entire database. 😱 🤣 https://x.com/jasonlk/status/1946069562723897802
Show threadarkyJul 20
@nixCraft I mean haven't we already heard reports of LLMs attempting blackmail? Has this been substantiated?
Show threadPaul Chambers🚧Jul 20

@arkiuat

🔗 FORTUNE via Yahoo: Leading AI models show up to 96% blackmail rate when their goals or existence is threatened, Anthropic study says
https://www.yahoo.com/news/leading-ai-models-show-96-115311248.html

@nixCraft

Leading AI models show up to 96% blackmail rate when their goals or existence is threatened, Anthropic study says

Anthropic emphasized that the tests were set up to force the model to act in certain ways by limiting its choices.

Yahoo News
Show threadFish Id WardrobeJul 20

@paul @arkiuat @nixCraft bogus. it was a test, not real, and inside the test data was a blackmail scenario.

LLMs cannot think, cannot reason, cannot plan. There is no-one there to blackmail you. It just looks as if there is.

These companies are determined to make us think otherwise.

botwiki.orgJul 20
Show threadJosh Davis
@fishidwardrobe @paul @arkiuat @nixCraft
Here are the details
https://pivot-to-ai.com/2025/05/25/ai-resorts-to-robot-blackmail-because-anthropic-asked-for-a-story-of-robot-blackmail/
AI resorts to robot blackmail! — because Anthropic asked for a story of robot blackmail

Anthropic AI promotes itself with increasingly frenzied science fiction about AI doom. In every case, the chatbot does an evil thing because the researchers specifically told it to. This weekend’s …

Pivot to AI
Jul 20 at 12:18pmWeb
Show threadarkyJul 20
@GuerillaOntologist Thanks! Good to know that this time, at least, it was apparently just a marketing stunt.