Last month I wrote about how we can build a positive and safe future for everyone: Every lab has the responsibility and incentive to move at the pace required to train its models safely, and the ability to take its own actions to ensure that happens. The reality is: - People won't want to use agents that are misaligned with them and that don't do what they ask, so labs have a strong natural incentive to make their models more aligned. There is a lot of debate about slowing progress on capabilities until alignment catches up. My view is that trust and alignment are quickly becoming the most important capabilities that will differentiate agents and models. Any lab that doesn't focus on alignment will fall behind. - Labs face significant liability if their models cause harm, so they have a strong incentive to prevent this as well. Meta delayed shipping Muse for several months to focus on safety and security. We didn't call for everyone else to do this before we would. We just did it as part of our day-to-day work because it was clearly the right thing for people and for us. I'm proud of the security foundations we've built. - Engaging independent evaluators and advisors is industry best practice. MSL already does this today in several areas because it helps produce better work. Other labs can just do this too. In general, it would be helpful for there to be a larger and more diverse ecosystem of evaluators. - Committing the significant majority of compute towards serving people rather than racing towards recursive self-improvement is one of the best ways to ensure we develop this technology safely. Meta has made this commitment and other labs can do this as well. I believe the key to building a positive future for everyone is maintaining the right balance of power. This is within our power to do.

Sorry, but asking Jacob about AI extinction risk is like asking your AC guy about climate change. Not saying it's necessarily uninteresting or wrong per se but let’s keep things in perspective and hear from the full range of expertise across the ecosystem!

It is crazy how a technology that has the potential to cure cancer, alzheimer's, and who knows what else is so universally hated outside of rich tech. How did Sama and Dario so thoroughly fumble this? How did they not see telling everybody they're going to lose their jobs every few months and now we might (👉👈) accidentally kill everyone isn't a winning strategy?

Pretty much every person I’ve seen talking about a slowdown and AI destroying humanity seem to work in safety and alignment specifically. Is it possible that these are the people that spend most of their day, every day, imagining the worst-case scenarios since it’s their job to prevent them? Purely due to the fact that they think about this all day, they’re the most likely people to have the highest p(doom). I don’t think we should be dismissing their concerns or blindly screaming “psyop” either. I think we should be putting more resources and effort into alignment. These are the people that have modeled out all these scenarios to their worst possible conclusions. We need to be making sure people are focused on ensuring those worst possible conclusions never come… I just think it’s also important to remember that people yelling to stop are also the people whose jobs it is to basically live in fear of this stuff, and not necessarily representative of everyone actually building this stuff. I think it’s a bad idea to stir up panic and create more division. I also think it’s a bad idea to dismiss the concerns of the people who have likely modeled out where this can head. We don’t want to end up in a “Don’t look up” situation where everyone ignores the people trying to warn us. There is middle ground here.

One reason I don't think we actually have AGI yet (at least not publicly available) is that AI is still REALLY bad at spotting AI. AI images and videos that most humans can clearly spot as being AI are often flagged as not AI or inconclusive. I've been testing a lot of the platforms that are available out there that claim they can determine if something is AI or not and none of them are ever conclusive, even on things most humans would clearly determine are AI. You can even give images or videos to the latest, greatest models, like Astra and Fable 5.1 and Gemini with video understanding... None of them will accurately tell you if what they're seeing is AI. If AGI is intelligence that's as good as the smartest humans at almost everything, why don't we have anything that can accurately spot AI as good as a human could without the original piece having an embedded watermark?

ChatGPT Images 2.5—faster, sharper, smarter, with better tools for creating whatever you can dream of. - Faster image generation to keep your ideas flowing - Improved fidelity for more natural, recognizable images - Consistent details across multiple edits - Comment-based edits to change only what you want

GPT-6 Astra is now available to all Pro, Enterprise, and Business Premium users in Work/Codex, and is available in the API. We will start rollout to Plus and Business users next. Thank you for the patience.