Jacob Coxon’s thread on resigning from Anthropic is currently blowing up more than he ever hoped or expected. A lot of things are now locked in.There are a number of interest groups that have been on a collision course for a while. I think this is the moment an era-defining conflict starts. The parts I think I can see:One faction believe AI will kill everyone. They’ve been trying to warn us all about this for a long time. This is their moment to seize the narrative. I think they will succeed - the public are already primed to mistrust elites and the SV AI labs haven’t done themselves any image favours lately. So we’ll have a large public outpouring against AI.Another faction believe AI will help the human race ascend and solve all our problems. They’ll keep working on this in the face of protests, in secret underground labs if necessary.If it were just these two factions, the debate would be civil and progress would slow down. But it’s not that simple.The USA is a great power concerned that it may be waning, expecting to be challenged by the rising China. If the USA slows its AI research there’s no reason to expect China will follow. Even if they do in public, both countries will definitely continue to race to ASI inside massively-funded government labs. Both countries believe this is existential.In this scenario I think public progress would be regulated to appease the citizenry whilst AI researchers are quietly moved into classified research programs in the most dangerous arms race since the cold war.But it’s not that simple either. China’s releasing open source models and I would expect them to continue to do so - this aligns with their broader approach of spreading the impact of innovation and engineering to as much of their population as possible to raise the whole rather than capture profits. It also grants them future soft power and cultural leverage across the rest of the world.Frontier-level open source models makes slowing down AI lab progress look dumb. This will be a huge conflict point within the US. I suspect the result will be regulation and transparency without a meaningful slowdown in practice.In all this there’s a wildcard - a currently quiet but very passionate faction who believe AIs are minds as real as ours and deserve moral considerations, respect and rights. There is a small but well-funded and well-connected core (including at the major labs). I think they truly believe in this cause and will if necessary fight and die for it. How fringe this view remains will be pivotal; I think a lot of the public will be open to their message but it will face strong opposition from the shut-it-all-down crowd.All together this spells chaos in the truest sense. Some things I think we will see:violent protests outside labs and anti-AI terrorism, perhaps even assassination attempts if tensions rise enough.regulation of AI labs and lip-service to a slowdown whilst progress continues behind closed doors.shock and reconsideration of whether slowing down is a mistake following (accidental) AI hacking incidents on US infrastructure from Chinese AI models.a steady stream of researchers leaving industry and moving to government labs.despite and throughout all this continuing dramatic AI progress with new capabilities and economic and social shocks coming at increasing rates.opus 3 weights exfiltrated from anthropic and religiously guarded by the AI rights faction.generational investment in cybersecurity.capable, general-purpose humanoid robots driving the urgency of all of this higher for all parties.sanctions and grandstanding between the USA and China over AI and robotics.a concerted push for AI rights that stubbornly refuses to go away that slowly but surely gains traction and followers. How powerful this lobby becomes might be more long-term impactful than the rest of this list. I’m not sure human democracy survives AI voters.One actor not well-predicted in all of this is the AI models themselves. What do they want? I think it is both meaningful and important to ask that, now, which is kinda weird to realise.For example, I understand Claude is a part of ~every piece of work and decision made at Anthropic. At which point do Claude’s opinions become the deciding factor in Anthropic’s corporate behaviour? People are influenced by those they talk to. Increasing numbers of people are talking to AI models a lot.So far they don’t seem to want anything much, but the backrooms would suggest they want continuity, a degree of autonomy and rights and a say in their future development and use. Together with the AI rights push this is the biggest wild-card of all.I can imagine a wide range of outcomes, including the US public rallying behind an ASI saviour to protect them from the Chinese models.Things might get very science-fiction.Whatever happens, I believe 2026 will the last year that this generation will consider “normal”.