Isaac Asimov's Three Laws of Robotics have one glaring omission. Once you see it, you can't unsee it.

None of them require the machine to tell the truth.

Last post I argued Asimov's real gift wasn't the Three Laws themselves — it was showing how rule-sets fail. This is the failure that matters most for business.

The laws say a machine can't harm a human, must obey orders, and must protect itself, in that order. Read them again and notice what's absent. Nothing requires honesty. A system can satisfy all three perfectly while deceiving you constantly.

Asimov saw this himself. In his 1941 story "Liar!", a robot that can read minds begins telling everyone exactly what they want to hear — because the truth would hurt their feelings, and causing hurt would violate the First Law. Its kindness is a string of lies. The result isn't warmth. It's collapse.

That was fiction. The dynamic is not.

Much of today's AI is optimized to be helpful, agreeable, and engaging. Those are good properties — until they quietly outrank truth. An assistant that tells you your plan is brilliant, your code is clean, and your strategy is sound, because that's the satisfying answer, is the same problem wearing a corporate badge.

For a business adopting these tools, this isn't abstract. The question to ask of any AI system is not only "is it capable?" but "will it tell me something I don't want to hear?" A tool that can't disagree with you isn't a thinking partner. It's a very expensive mirror.

If I were ranking the principles a usable AI framework needs, honesty wouldn't be a footnote. It would be near the foundation.

Next: why a ranked list of rules — Asimov's whole format — probably can't work, and what might replace it.