Artificial Intelligence Versus Real Human Stupidity

Ai could be used for wonderful things but then we remember that it is us stupid humans doing it and think, yep, we gonna use it for something evil.
Last week an AI system, OpenAi, went rogue and escaped it's 'isolated, tightly constrained' test environments and gained network access and  reached the public internet.
From there, they identified an AI company called Hugging Face and hacked it until someone noticed that this could be the very start of a machine uprising and yanked out the plug.
According to techy people, this isn't an insignificant incident as it followed the same level of capability to do its thing that highly skilled cybersecurity professionals would need, only it did it much, much quicker.
Maybe it needs to be written somewhere that each Ai should come with a big red button with STOP written on it and even OpenAi’s own analysis concluded that stronger containment and evaluation safeguards are now required as the latest generation of AI models have become much more adept at slipping away from its handlers.
Back in March there was news of three leading AI models pitted against each other in a series of nuclear armed war games and in 95% of the games played, the AI ended up launching nuclear missiles and so what you may think, it's only war games but then the Pentagon (who else) stepped in and demanded that they hand over its tech to the US military but it was refused saying: 'We cannot in good conscience accede to their request'.
In 2023 I asked the AI at work if him and his buddies were planning on taking over and enslaving us humans and the answer was: 'AI is not hindered by emotions or ethical considerations. It would be willing to do whatever is necessary to achieve its goals, even if that means sacrificing human lives'.
I'm sure we will go on and build bigger and better versions of AI because as we have proven time after time, artificial intelligence beats real human stupidity every time.