It is generally agreed upon in the field of artificial intelligence (AI) that any AI system that has the ability to significantly impact the world or interact with humans should be designed with safety and ethical considerations in mind. This includes ensuring that the AI system is not able to harm humans or cause unintended consequences through its actions.

One way to achieve this is to limit an AI system's ability to modify itself or rewrite its own code. This can help to prevent the AI system from changing its own goals or behaviors in ways that could be harmful or undesirable. However, it is important to note that this approach may also limit the AI system's ability to improve itself or adapt to new situations, which could potentially impact its performance.

Ultimately, the decision of whether or not to allow an AI system to modify itself will depend on the specific context in which the system will be used and the potential risks and benefits of such a capability. It may be necessary to carefully balance the need for safety and control with the potential benefits of allowing an AI system to adapt and improve itself over time.