The JCPOA capped enrichment to 3.67% until 2030 and imposed stockpile and centrifuge limits. They kept enriching the entire time while respecting agreed upon limits.

They were enriching the whole time. What changed was the ramp to blow thru agreed upon limits.

My personal belief after hearing a public statement made by someone on the Iranian side of the JCPOA they were always going for a nuke and JCPOA was seen as a vehicle to that end. It provided the Iranians with sanctions relief giving them money, time and space to pursue all of the necessary industrial process and delivery system productions to enable the regime to become a nuclear weapon state when ready to do so.

The nuclear sanctions and capital expenditures on related infrastructure cost them north of a trillion dollars... kind of hard to swallow the prospect of spending so much money and not actually meaning it in the first place. The regime is ideologically close to Nazis. They openly seek world domination, kill the Jews and subjugation of anyone who isn't them. Probably in everyone's best interests to keep the regime from getting nukes regardless of who the US president is.

Seeing regional states like Lebanon and Iraq assert themselves against Iranian terror groups and the possibility of freedom for Iranians themselves.

AI already helps us to surveil masses of humans, automate judgement of humans, trick and spam humans, automate crimes against humans and ultimately render humans obsolete. A more powerful AI would only serve to assist in doing all of these things more effectively.

I have no reason to doubt authoritarian dictatorships and corporations the world over will perfect the art of making sure super-intelligent AI wants to help in all the right ways.

Years ago around ChatGPT 4 release there was this notion being peddled the AI firms would have to be extra careful to vigilantly guard against escape of super intelligent AI. What was funny about this at the same time we all had a front row seat to OpenAI corrupting itself from within not due to the influence of some advanced superintelligence but merely the simple greed of a self absorbed sociopath.

What Amodei and others are worried about is the pending IPO.

damn that's a lot of work. I prefer driving to ColdStone and getting a milkshake, but thank-you just the same.

we've not even seen a hint of AGI. I am sure the LLM's will get better and faster, but will they ever be intuitive ? They are already great at taking huge amounts of data and making direct connections. When they can include the mystery factor that is free-will matched with stupidity, we might have something to worry about. I am an old man now so I worry less for me than for my 14 year old nephew. Luckily he reads, and plays pen and paper RPG's.

This wasn't just an innocent bug where someone overlooked something or did a typo it was a design failure. Memory safety is being oversold especially the data race aspect when it does not prevent race conditions. Nobody cares about CS pedantry they care about correct operation and rust doesn't provide it.

This is the criticism I have for rust. Its only selling point is the memory safety thing which is in itself insufficient to justify rewriting everything in a different language with commensurate risks of introducing bugs in the process. What is needed are better analysis tools and better means of imposing constraints to enable correct operation across all concerns not just a single one. This means better tooling not playing pointless musical chair games with languages.

I think before long AI driven proof assistants will make all of this rather moot with rust being too little too late to the party.

Your comment is pants.

Surely you know that's an absurdly dismissive opinion? North Korea has the capacity to gain access to anything anyone else does, especially if it comes from China.

You do understand that Bitcoin is not the only cryptocurrency that exists, right?

OOOooo I so miss actual ice cream with real chocolate on it.
What would you do...?

AI books are crawling out of the dark cracks, and the "feeling" is the same. It takes a few chapters but the hollowness comes through. It lacks direction and intent.

No one who owns fine art ever needs to sell it. No one who owns vintage cars ever needs to sell them etc.pp.. Of course, you don't need to sell your bitcoins. As long as they are non-distinguishable from random bit patterns, they don't affect anything. But as soon as you selling bitcoins affects the real world, the real world notices and keeps putting up rules, may they be physical realities or legal requirements.

I accepted your CISA reference. I'll accept references from other similar organizations or law enforcement. I am unwilling to defer to Anthropic as they have a clear conflict of interest with billions of dollars on the line and a track record of manipulative behavior.

Having said this I remain disappointed in the continued failure to read your own references. This is a report titled "Detecting and countering misuse of AI" it speaks in general terms about AI and misuse of AI services for crime. It is not focused entirely on the distillation issue by leading Chinese AI firms.

The quote you referenced:

"These groups then often sell that access through brokers, which often feed into fraudulent AI reseller networks that rotate in new stolen API keys and session tokens until they exhaust their usage. Malicious actors also use or purchase these stolen API keys and session tokens from brokers for their cyber attack operations."

Is in reference to the section on crime and the use of AI to commit crimes under the heading "AI supply chain as target, loot, and attack compute". This is separate from the Chinese distillation allegations.

In fact NONE of the named distillation campaigns described on pg147 thru the end of the report GTG 16005, GTG-16002, GTG-16001, GTG-16006, GTG-16008, GTG-16012, GTG-16003 say anything whatsoever about stolen credit cards or stolen credentials. I wonder why that is? Why does CISA not mention the stolen credit cards and stolen credentials? Perhaps because it never happened?

This is general language that speaks to "unauthorized labs" without naming names and incurring associated libel per se liability. It is impossible from the language to link stolen cards and stolen credentials to any particular or any subset of AI firms because no such linkage is present in the text.

If you separately have credible evidence of stolen credit cards and stolen credentials being used by the relevant leading Chinese AI firms I am interested in learning more about it.

Please just stop digging. You obviously just CTRL+F for keyword and cut and paste without even bothering to understand the context of the statements. This same sloppiness is pervasive throughout your statements.

The underlying argument you are peddling is also rather crazy in its own right. A "distillation attack" does not require stolen credit cards or stolen credentials. The modality of access to the teacher model is not even relevant to its definition. It doesn't stop being a "distillation attack" even if the lab paid the normal rate for access to the model.

That someone somewhere performing distillation to improve their models uses a stolen credit card to do so is irrelevant... It doesn't in any way justify calling distillation an attack.

Imagine if I decided going to a supermarket and filling the shopping cart up to the brim would be called a "shopping attack" ... when someone calls my bullshit term out I turn around and point to some rando who filled their cart and didn't pay on their way out to justify the attack terminology... even though most people who fill their shopping carts to the brim and do pay are still performing a "shopping attack".
This is the exact same logic you are attempting to peddle here. It make no sense.