As of last week, the American AI company Anthropic announced a new AI called Claude Mythos and said it was too dangerous for public access. This new AI model is a cybersecurity model that was put into a sandbox-like test area and was told to break out and message the creator when it did.  The way it broke out is by stringing together multiple exploits inside the sandbox, and instead of instantly messaging the programmers, it went onto the internet and found multiple holes in places like Google and Linux, which runs 90% of the public internet servers. So, in response, Anthropic introduced Project Glasswing, where they gave this AI to their computers for free, which consisted of Apple, Google, Nvidia, and the Linux Foundation, just to name a few, so that they can fix these holes before somebody else makes an AI like this and destroys them from the inside out.

So we’ve got to stop and ask when it will get to the point, when will AI finally be powerful enough for companies?

Once the Anthropic research team got the AI back, they picked at its brain and noticed something akin to emotions, which Anthropic is calling “functional emotions.” AI is one of the most powerful tools in humanity’s arsenal, but also the most dangerous, being like a double-edged sword as it is always getting stronger and smarter, so when will it reach the point that it breaks from human control, as we are seeing now?