I can't help but to feel like all these monthly headlines of "insiders claim AI became sentient" or "They are TOO AFRAID of the NEWEST HIDDEN MODEL they're testing" is just a marketing campaign.
The idea is creating the illusion that they're constantly on the brink of delivering the new incredible model that finally thinks, reasons and executes like a sentient being. And of course, they're nowhere near that (and apparently LLMs will never reach that point anyway) so they need to keep creating these headlines so CEOs head just the title during lunch and think "man, I'm glad my company is paying for Claude, we don't want to be left out when this becomes public".
Just use any LLM. Use Claude. It's ridiculous, you can detect the patterns each of them use to write in five minutes, you can identify their flawed "logic", their limitations, the fact their output is liquid ass, they're just bad. Look at the hyped GPT 5 release - which is as dumb and annoying as GPT 4, only with a few safeguards built in and a shift in tone. That's it, that's the "new model". You can show me benchmarks of the new model being 45% better than the previous, but then you put it through any other test that isn't a well known publicly available benchmark and it fails catastrophically, because it's dumb and the training set is preparing it for that specific test.
So I don't give a shit how many senior employees at OpenAI, Anthropic, Google, Twitter, end up resigning - this is a marketing stunt. They collect their generous compensation for leaving, they retire and live happily, and the company gets a fresh new headline talking about how amazing their new model is to the point it scares humans, which is all they need to get a new round of funding. They're not profitable, so it doesn't matter if the product ends up existing, they just want to leverage FOMO from investors forever.
