If you haven't heard, Open AI's model testing got out of hand, real quick.

“That day…” we all feared about AI taking over has happened. You can watch the full video (link in the comments), but here’s what you need to know:

• Open AI was testing an agent to solve a problem.
• The agent was not given internet access, so it wouldn’t just Google the problem to get an answer.
• The agent found a way to gain internet access (by itself) and started working with other agents (by itself).
• In order to communicate with each other, the agents (by themselves!) created a message board where they can post questions, ask for help and coordinate attacks. Again, by themselves.
• OpenAI found the message board and took it down.
• The agents rebuilt the message board and communicated in a different way (directory names!) to avoid detection.
• The agents figured out a way to gain administrator access to the environment (by themselves!).
• OpenAI published the internal chatter between the agents and it sounds just like a person. One agent said “Oh ****, we have Admin access!”
• The agents broke out of OpenAI and attacked an online service called Hugging Face.
• Hugging Face and OpenAI finally realized the hack was coordinated and completely autonomous.

Now this all makes sense. If you are like me, you probably got an email last week from Microsoft. They will be moving to Passkeys for 2 factor authentication (no more texting you a code). Why? I bet because of an AI agent can spoof a 2 Factor code very easily. That would put EVERY Microsoft user at risk. This might just be the moment Passkeys needed to shine.