And I'm sure we've all seen what happens if you run the Declaration of Independence or the Gettysburg Address or the book of Genesis through an AI "detector". They usually come back as AI.
Only for poor quality systems. Unfortunately there are many systems that tried to make easy hype, but are the equivalent of an ML 101 classifier class project.
If one measures for perplexity (how likely text is under a certain language model), common text in a training set will be very likely. But you can easily create better models.
At least he owns his own domain and can eventually switch over. A few years ago we decided to switch our personal emails from gmail accounts to domains we own (though the email is still handled by google.) This way if we ever lose our google account, we can switch the MX and be able to get all our recovery emails, bank second factors, password recoveries, etc.
They could switch their domain to another email provider and start getting emails, which is great. The problem though, is they also used their Google Account to log in to all the 3rd party services (payroll). I have no idea how you would get back into those services. Some _might_ let you switch off the Google Sign-in SSO, but I imagine that is a headache.
I don't know if you're kidding, but I agree with you. I use the Copilot CLI in VS Code and Visual Studio and it works better than anything else. I do use Claude models with it....
These drive my password manager nuts, especially when there are actually different logins for them. I just put a note in it saying exactly what service it's for.
However, the numbers stations transmissions are never a big secret. They're intentionally powerful so someone can pick them up on simple equipment without raising suspicion. A person can modify an off-the-shelf AM radio to pick up shortwave, for example, even in an oppressive regime.
It's a one-time pad, so the encryption is unbreakable.
reply