Curious about this lawsuit - I assume that there are legal means to “collude” on safety. Car companies coming together to agree on crash test regulations will “slow down” auto manufacturing but I’m pretty sure this mechanism is legal.
If the four companies came together to start an industry organization and agree on safety protocols, I assume that would be legal done correctly.
Serves them right, it was obvious BS, likely cause the return on investment for frontier training is not there anymore. They didn't think through the implications of the "We're responsible stewards now" PR.
I don't know that I know that for sure. But I also won't be surprised.
One way that I think about it is to look at what's happening in a completely different regime to see how they approach AI safety. So looking at China, how do they balance safety against freedom to develop the technology and how does it compare to our (US) approach?
(I think of China, not Europe, because there's not much AI innovation coming out of Europe so they don't need to navigate the tension to same degree as the US and China).
From the government side, the approach is mostly to monitor the situation and deal with risks as they arise, e.g. version 3.0 of the AI Safety Governance Framework https://www.tc260.org.cn/tc260/xwdt1/202609/e879077a3caa4722... was just released this month, now also covering "unintended autonomous behaviors."
Curious about this lawsuit - I assume that there are legal means to “collude” on safety. Car companies coming together to agree on crash test regulations will “slow down” auto manufacturing but I’m pretty sure this mechanism is legal.
If the four companies came together to start an industry organization and agree on safety protocols, I assume that would be legal done correctly.
Serves them right, it was obvious BS, likely cause the return on investment for frontier training is not there anymore. They didn't think through the implications of the "We're responsible stewards now" PR.
> it was obvious BS
I don't know that I know that for sure. But I also won't be surprised.
One way that I think about it is to look at what's happening in a completely different regime to see how they approach AI safety. So looking at China, how do they balance safety against freedom to develop the technology and how does it compare to our (US) approach?
(I think of China, not Europe, because there's not much AI innovation coming out of Europe so they don't need to navigate the tension to same degree as the US and China).
On the industry self-regulation front in China, the Artificial Intelligence Industry Alliance came together to formulate a set of safety commitments https://aihub.caict.ac.cn/files/aiia_security/content.pdf in 2024, though my impression is that it was more of a one-time thing, since for example the associated webpage http://web.archive.org/web/20250813202126/https://aihub.caic... now 404s.
From the government side, the approach is mostly to monitor the situation and deal with risks as they arise, e.g. version 3.0 of the AI Safety Governance Framework https://www.tc260.org.cn/tc260/xwdt1/202609/e879077a3caa4722... was just released this month, now also covering "unintended autonomous behaviors."