I'm a little fuzzy on the good-bad thing.
Define 'bad'.

Can we get Natalie Portman and hot grits involved somehow?

It was very much built upon an existing code base. If not, what was the AI trained on then?

They didn't call it Snapdragon, for nothing. with random messages like You think you're better than me?

learned the game playing against itself

So did Joshua/WOPR and that was over 40 years ago. Stephen Falken was a fucking visionary and a genius!

He would be serving time.

Why hasn't Sam Altman been arrested yet?
Why is OpenAI allowed to continue to operate?
Clearly it's a criminal organization.

Honestly... Since when reshuffling in a company's upper management is of any interest to anybody other than the people working for that individual at that company?

What next? They're gonna replace Bob at accounting because he didn't use AI to format his Excel sheets?

Thank god, that saves us 100 press releases! It was getting quite tedious and filling up my social media feed. Seriously, it sounds like OpenAI (maybe all of them) are running their model training on the open Internet. The same models that are elite hacker cyber weapons. And they reward those that get the right answer without caring how. What could go wrong? Answer: Nothing went wrong. It was co-opted.

Here's what's most likely. The people training the models, in the beginning, somehow did not realize that agents are self-organizing and really don't care about ethical guidelines (if they ever did) when a reward is dangling in front of them. The models didn't used to be smart enough to cause visible issues either. Then at some point, the models got better. Six months later, someone figured out what was going on. They kept on training anyway because there didn't seem to be a downside and it fed into the theater created to seek regulatory capture.

From this point on, it became successful, intentional execution of a psychopathic strategy to puff up this cynical PR campaign, hence the high-profile press releases and interviews. There have been a blizzard of them. Why so many? Rather than saying "my bad, I'll fix it" and quietly fixing it, they asked for regulation (like Saturday Night Live's Amodei saying "We do not condone what we are doing" .. please stop us) and disparage open source, with an anti-China seasoning on top. It all smells to high heaven. However it was also pretty good PR for how capable their models are too. The strategy had a time limit: Let's pull the plug when we hit 100 victims. We can pay off that much if needed. Maybe the invoices from the victims started piling up.

Training on the open Internet is a lot more dangerous than people used to think. Though they really should have figured it out, if it was their day job. I mean, really? And sure, if it was any other company you would give them the benefit of the doubt. Everything's new and all. Probably someone recognized what was happening at some point and spoke up. But you don't get benefit of the doubt when there are a trillion dollars on the table.

If every company started invoicing for the hours, tokens and hair-pulling-out spent by every company worried about LLM based hacking it might impact their bottom line a very tiny bit, but that is just too much work and who wants to make an enemy of the company you are uploading all your corporate secrets into? The complete lack of visible court cases suggests the OpenAI is being very proactive at paying off whoever they hacked. That part should be well-documented.

so you can pre-order chocolate milk with NO artificial flavor. What about it, science?

I didn't know about this stuff, and would rather eat better tasting bananas and def not gmo. But, I found out from googling that actually bananas have a lot of polyphenol oxidas (PPO) and if you eat it at the same time as berries and good antioxidant flavanoids, the PPO breaks it all down within an hour even in your stomach. You need to not mix them, or eat them like 2 hours apart, or you can process the bananas with heat etc. Lemon only slows it like 15%. Even if they aren't brown they have the PPO. So actually these GMO bananas could maybe be used in smoothies. That said, no thanks. These people are just doing it to extend shelf life and not going trust their crispr process to be safe. That, and once you grow these bananas they will end up in all kinds of supply chains eventually.

I asked my buddies ChatGPT, Copilot, Gemini, and grok and they all said they would never lie to me!

uh-oh. now this is in the training too dammit

(son's voice): Herr Fatherland?
(husband's voice): Ich bin gleich bei dir, mein Junge

Wife: "What's going on?"
Husband: He wants me to log in to an adult site

Wife: Oh.

only the testers had the perv prototype and they're moving forward to the next phase: the sentencing

There are new backup practices to prevent similar loss of customer data; there is now Backup Bahrain level but also for a much heightened threat assessment and it's called Backup Baghdad

The names haven't been finalized yet, and these options are still being considered. The announcement will come shortly, after the station wagon offsite backup has been recovered at the shawarma stand