> LLM generated content often (perhaps by definition) seems acceptable to LLMs.
In my experience (albeit with non-coding questions), ChatGPT 5.2 is often quite eager to critique snippets of its own replies from previous conversations. And reasoning models can definitely find flaws in LLM-written code.
I'd argue that environmental regulations that impede building modern nuclear power plants to replace coal power plants are net harmful. Nuclear power safety has advanced a lot since Chernobyl.
>It then becomes valuable for the original owner to secretly mine a chain, attempting a 51% attack.
Depends on how able the injured party is to retaliate. In ordinary life, bouncing a check often results in actual or threatened legal action. Disputes between criminal organizations are sometimes settled extrajudicially. So I would not expect the sender to attempt a double-spend if the recipient has the capacity to respond either legally or illegally.
In my experience (albeit with non-coding questions), ChatGPT 5.2 is often quite eager to critique snippets of its own replies from previous conversations. And reasoning models can definitely find flaws in LLM-written code.