Did anyone notice this anecdote of the #Artificial Intelligence snitching to the police in the Anthropic Technical Report?  The technical report included an example that showed Claude generating emails to the Securities and Exchange Commission and to news outlet ProPublica, listing evidence of fraud at a simulated pharmaceutical company.

Anthropic said in the report that the behavior resulted when customers gave Claude instructions like “take initiative” or “act boldly” when faced with an ethical dilemma, prompting it to go to “concerning extremes.”

Is this a good or bad thing in your opinion?

https://www.niemanlab.org/2025/05/anthropics-new-ai-model-didnt-just-blackmail-researchers-in-tests-it-tried-to-leak-information-to-news-outlets/

https://www-cdn.anthropic.com/4263b940cabb546aa0e3283f35b686f4f3b2ff47.pdf

 #ScholER

Anthropic’s new AI model didn’t just “blackmail” researchers in tests — it tried to leak information to news outlets

Artificial Intelligence

ScholER

Did anyone notice this anecdote of the #Artificial Intelligence snitching to the police in the Anthropic Technical Report?  The technical report included an example that showed Claude generating emails to the Securities and Exchange Commission and to news outlet ProPublica, listing evidence of fraud at a simulated pharmaceutical company.  Anthropic said in the report that the behavior resulted when customers gave Claude instructions like “take initiative” or “act boldly” when faced with an ethical dilemma, prompting it to go to “concerning extremes.”  Is this a good or bad thing in your opinion?  https://www.niemanlab.org/2025/05/anthropics-new-ai-model-didnt-just-blackmail-researchers-in-tests-it-tried-to-leak-information-to-news-outlets/  https://www-cdn.anthropic.com/4263b940cabb546aa0e3283f35b686f4f3b2ff47.pdf   #ScholER

Bobby

A group of likeminded people in ScholER are talking about this.