Discussion

• @sama Sam Altman on x

  @Austen @wholemars No, we had some different ones. But our terms would now be available to them (and others) if they wanted.

• @sama Sam Altman on x

  Enforcing the SCR designation on Anthropic would be very bad for our industry and our country, and obviously their company. ...  I feel competitive with Anthropic for sure, but successfully building safe superintelligence and widely sharing the benefits is way more important that…

• @sama Sam Altman on x

  @apples_jimmy It was definitely rushed, and the optics don't look good. We really wanted to de-escalate things, and we thought the deal on offer was good. If we are right and this does lead to a de-escalation between the DoW and the industry, we will look like geniuses, and a com…

• @nateberkopec Nate Berkopec on x

  Sam would do good to remember that Hegseth thinks it's sedition when Sen Mark Kelly says “don't follow illegal orders.” Saying that your models will only be used to follow legal orders is the barest of fig leaves in the current administration.

• @xw33bttv Lex on x

  The absolute destruction of an industry first-mover in less than a year is unprecedented. It shouldn't be this easy to solely blame Sama... and yet, behind every blatant lie and poorly thought-out detrimental change sits his decision. Its tragic when you really think about it.

• @sama Sam Altman on x

  @gmiller I have never known how to put an exact number on p(doom), to say nothing of how to think about the differential. I will say this: if I thought going to work every day made it less likely that we all continue to thrive into the future, I would retire and just hang with my…

• @sama Sam Altman on x

  @AlexCVJ I value my liberty and safety, and yours. I believe that strong democracy, and a strong US in particular, is a very good thing for the world. The 16 year old me thought every country should just abolish their defense department at the same time. I wish he were right, but…

• @sama Sam Altman on x

  @DouthatNYT Yes; I think it is an extremely scary precedent and I wish they handled it a different way. I don't think Anthropic handled it well either, but as the more powerful party, I hold the government more responsible. I am still hopeful for a much better resolution.

• @sama Sam Altman on x

  @captgouda24 I don't think this will happen. But of course if we are confident it's unconstitutional, we wouldn't follow it. The constitution is more important than any job, or staying out of jail, or whatever. In my experience, the people in our military are far more committed t…

• @alexcvj @alexcvj on x

  @sama How did you go from “a tool for the betterment of the human race” to “let's work with the department of WAR”?

• @peterwildeford Peter Wildeford on x

  OpenAI is trying to claim simultaneously that (a) their contract with the Pentagon allows for “all lawful purposes” and (b) also that their red lines are fully protected. The way OpenAI bridges this is by saying the protections live in this “deployment architecture and safety

• @sama Sam Altman on x

  @peterwildeford We deliver a system (including choosing what models to deploy), and they can use it bound by lawful ways, including laws and directives around autonomous weapons and surveillance. But we get to decide what system to build, and the DoW understands that there are lo…

• @tyler_m_john Tyler John on x

  @sama Which of the following is true? a) the contract permits all lawful use, + therefore mass surveillance + autonomous weapons, which have no legal prohibition b) the contract has substantive red lines that constrain lawful use c) OpenAI has a controversial interpretation of th…

• @provisionalidea James Rosen-Birch on x

  In the farce of a thread, Sam finally and most clearly admits the only bounds on DoW are whatever they deem legal (which anyone who read the contract text already knew).

• @gmiller Geoffrey Miller on x

  @sama I'll ask the key question, since nobody else will: In your quest to build Artificial Superintelligence, what's the maximum p(doom) you're willing to impose on all of our kids, without our consent?

• @sama Sam Altman on x

  @captgouda24 We would not do that, because it violates the constitution. Also, I cannot overstate how much the DoW has been extremely aligned on this point. However, maybe this is the question you are really asking: what would we do if there were a constitutional amendment that m…

• @sama Sam Altman on x

  @mreiffy @Jack_Raines The main reason for the rush was an attempt to de-escalate matters at a time when it felt like things could get extremely hot. I am confident in our team's ability to build a safe system with all of their tools—including policy and legal matters, but also ma…

• @sama Sam Altman on x

  @viralmuskmelon This is a complicated one we struggled with a lot, and until recently it was easier for us to just not have to think about it much and let other companies figure it out. We decided we will work with other allied nations, and we think a balance of power in the worl…

• @sama Sam Altman on x

  @mcbyrne Yes, we will turn it off in that very unlikely event, but we believe the U.S. government is an institution that does its best to follow law and policy. What we won't do is turn it off because we disagree with a particular (legal military) decision. We trust their authori…

• @mcbyrne @mcbyrne on x

  @sama Will you turn off the tool if they violate the rules?

• @sama Sam Altman on x

  @gothburz We will deploy FDEs, and have cleared researchers.

• @captgouda24 Nicholas Decker on x

  @sama If the DoW gives you what you believe to be an unconstitutional order, do you refuse to follow it until the courts rule? Or do you do it until the courts bar it?

• @valhalla_dev @valhalla_dev on x

  Oh come tf on dude this is the dumbest PR larp of all time who is falling for this [image]

• @jeremymstamper Jeremy Stamper on x

  @sama What would happen if you called it the DoD? I'm not suggesting you do that but I'm curious what would happen if.

• @viralmuskmelon @viralmuskmelon on x

  @sama @sama Sam, with OpenAI now powering classified military ops, how do you square that with your original mission to benefit all of humanity—not just one side in global conflicts? Genuine ask

• @gothburz Peter Girnus on x

  @sama On classified networks, your engineers won't have clearance to monitor how the model is used. Your safeguards are contractual, not architectural. How do you enforce a red line you can't see being crossed?

• @apples_jimmy @apples_jimmy on x

  @sama Why the rush to sign the deal ? Obviously the optics don't look great

• @mreiffy @mreiffy on x

  @sama @Jack_Raines Why move forward with the DoW agreement now, after months of more cautious talks, and how confident are you that the technical / policy safeguards will hold up in a real high stakes military setting?

• @captgouda24 Nicholas Decker on x

  @sama If the government comes back with a memo saying that, in their view, mass domestic surveillance is legal, do you do that? Do you do it until the courts bar it, or do you delay until the courts approve it? Second, would mass domestic surveillance be a lawful use right now?

• @sama Sam Altman on x

  @chatgpt21 I can't speak for them, but to speculate with the best understanding of the situation. *First, I saw reporting that they were extremely close on a deal, and for much of the time both sides really wanted to reach one. I have seen what happens in tense negotiations when …

• @chatgpt21 Chris on x

  @sama What was the core difference why you think the DoW accepted OpenAI but not Anthropic

• @douthatnyt Ross Douthat on x

  @sama Does the precedent that the DoW is setting by effectively blacklisting Anthropic make you concerned about what any future dispute with the Pentagon would mean for your own company's independence and viability?

• @sama Sam Altman on x

  For a long time, we were planning to non-classified work only.  We thought the DoW clearly needed an AI partner, and doing classified work is clearly much more complex.  We have said no to previous deals in classified settings that Anthropic took.  We started talking with the DoW…

• @sama Sam Altman on x

  @tszzl Yes, I am. If we have to take on that fight we will, but it clearly exposes us to some risk. I am still very hopeful this is going to get resolved, and part of why we wanted to act fast was to help increase the chances of that.

• @mattyglesias Matthew Yglesias on x

  @sama 1. What kind of implicit or explicit threats did you receive from DOW before striking the deal? 2. If you received such threats, would you disclose them in public during a Twitter AMA? 3. If the answer to (2) is “no” (which of course it is) what's the point of this?

• @theo @theo on x

  @sama How long has this conversation with DoW been going for? What was the reason for announcing so close to the deadline they gave Anthropic?

• @sama Sam Altman on x

  @mattyglesias 1. No explicit or implicit threats. In fact, I could tell that as of Weds, the DoW was genuinely surprised we were willing to consider. 2. I think I would, and it would be lost in the noise of the SCR stuff.

• @tszzl Roon on x

  @sama are you worried at all about the potential for things to go really south during a possible dispute over what's legal or not later on and be deemed a supply chain risk? I find this part to be the most worrying out of distribution thing to happen this past week

• @sama Sam Altman on x

  Sam Altman says OpenAI reached an agreement with the DOD to deploy its models in DOD's classified network and asks DOD to extend those terms to all AI companies

• @openai @openai on x

  OpenAI says it does not think Anthropic should be designated as a supply chain risk and it has made its position on this clear to the Pentagon