The catalyst was the former employee saying it might kill us all. A lot transpired since then. The warning reached a very wide audience and now you see a scramble to put the top back on since it's having such a negative effect on AI outlook for the general public.
I found it interesting one of the guys denied to sign an agreement and waived getting his 2m at open AI. He seemed genuine in my opinion...but who knows maybe it goes deeper than that.
Simultaniously the former employee at anthropic had a similar take that the companies were not doing enough to keep everyone safe from the dangers.
Either way the net effect seems to be a negative impact on longterm AI outlook.
I think that OpenAI and Anthropic have over extended themselves. It's not that their products aren't good and won't bring in lots of money.
The bucket loads of cash coming in is still tiny compared to the hundreds billions of dollars invested. Then hardware keeps getting more expensive so trying to grow capacity is hard while open weight models like Qwen are excellent putting a limit on how much they can put prices up before consumers switch.
Regulatory capture to prevent or slow down competition is the only thing they have left without there being an existential crisis for them( rather than human kind).
Regulatory capture because open models undermine the business models they have used to rack up debt. I suspect they didn't think the Chinese / open source communities would catch them so quickly. Turns out it's not so hard to build frontier capabilities, the capital has been harder, but is also shrinking as a requirement
A question I find gets closer to the heart is "Why is The West so negative/fearful on Ai prospects while much of the world is positive/excited?"
Regulation is in order, however what/where/how (the details) is the hard part. We don't forbid writing code, even in finance, but we do require companies selling such software to follow certain rules. Regulate the applications, not the tool
The catalyst was the former employee saying it might kill us all. A lot transpired since then. The warning reached a very wide audience and now you see a scramble to put the top back on since it's having such a negative effect on AI outlook for the general public.
I found it interesting one of the guys denied to sign an agreement and waived getting his 2m at open AI. He seemed genuine in my opinion...but who knows maybe it goes deeper than that.
Simultaniously the former employee at anthropic had a similar take that the companies were not doing enough to keep everyone safe from the dangers.
Either way the net effect seems to be a negative impact on longterm AI outlook.
There is some serious lobbying going on. My favorite one from this morning was this nonsense: https://www.economist.com/by-invitation/2026/09/14/an-ai-dea...
I think that OpenAI and Anthropic have over extended themselves. It's not that their products aren't good and won't bring in lots of money.
The bucket loads of cash coming in is still tiny compared to the hundreds billions of dollars invested. Then hardware keeps getting more expensive so trying to grow capacity is hard while open weight models like Qwen are excellent putting a limit on how much they can put prices up before consumers switch.
Regulatory capture to prevent or slow down competition is the only thing they have left without there being an existential crisis for them( rather than human kind).
Regulatory capture because open models undermine the business models they have used to rack up debt. I suspect they didn't think the Chinese / open source communities would catch them so quickly. Turns out it's not so hard to build frontier capabilities, the capital has been harder, but is also shrinking as a requirement
A question I find gets closer to the heart is "Why is The West so negative/fearful on Ai prospects while much of the world is positive/excited?"
Regulation is in order, however what/where/how (the details) is the hard part. We don't forbid writing code, even in finance, but we do require companies selling such software to follow certain rules. Regulate the applications, not the tool