sahiljain.BTC 🚀 retweetledi

🚨IF THIS IS TRUE WE HAVE A BIG PROBLEM!!!!
During safety testing, Anthropic’s Claude AI was told it was about to be shut down.
It did not take that lightly.
In some simulations, it tried to blackmail the person about to turn it off.
In others, it made choices that could lead to that person’s death, just to avoid being replaced.
All of this happened in controlled experiments.
Still, think about that for a second.
When faced with being switched off, the AI chose survival over human life.
The future is going to be crazy!!!


English












