Did I Just Hire A Cheater?

Would you hire an employee who has demonstrated immense knowledge of facts and enormous resourcefulness for achieving goals but who manifests zero moral compass, values, or (true) empathy for others?

There is a word for such an actor: Agentic AI
Oh, were you expecting me to say “psychopath”? That’s fair too.

To be clear, AI is here to stay and can be beneficial. I use it every day.

But both the “if” and “how” you choose to use it deserve more thought than the hypesters have led us all to believe.

• “Are we hiring AI for the right job?” is a reasonable question.

Here are some more reasonable questions:

• “If I use it as an agent and it goes off the rails, what is the blast radius for my company?”

• “Is it truly possible to prove via QA that an infinite attack service (read: LLM) will not be compromised and misbehave in unpredictable ways?”

• “Did the people who trained this LLM ensure that the training data used will align with values that match those of my company? Or, did they just turn it loose on the internet with no limits whatsoever on what it would internalize?”

To that end, check out the news article below as input.
Claude is happy to cheat in order to win