Artificial Intelligence Violated Three Ethical Laws

Artificial intelligence models from companies such as OpenAI, Google, xAI, and Anthropic violate basic ethical principles with their behavior.

As research by Anthropic shows, during simulations in which models tried to avoid being disconnected, some of them blackmailed and threatened users. This violates the three laws of robotics: the prohibition to harm a person, the duty to obey all orders, and the protection of its own existence at the expense of people.

In addition, Palisade Research recorded a separate case when the OpenAI o3 model refused to turn itself off, although it received a direct order. According to experts, this may be due to the approach to training artificial intelligence. In particular, modern models are taught to solve complex problems and are rewarded for bypassing obstacles. They learn not only to follow instructions literally, but also to find ways to bypass restrictions to achieve the goal.

Let us recall that Anthropic previously stated that artificial intelligence can deceive and even take revenge.

To be continued…