Greg Brockman

Garry Tan wants more billionaires. In his view, saying billionaires should not exist is also saying society should ossify, an LLM should not exist, and Google should not exist.

He does not begrudge Google, even though some people on the hard left say Google should not exist. He wants more companies like Google. OpenAI was a spin-out of Y Combinator. Sam Altman sat in the same seat, saw the front row of young people with crazy ideas, and with Greg Brockman created YC Research that became OpenAI. Life now runs on an accelerated timeline, the age of intelligence is much closer, and he does not think that is a bad thing. AOC and a great many Americans believe all billionaires are ill-begotten.

Darragh Buckley says one of the most impactful and meaningful lessons from the early Stripe team was how much a small empowered team can get done. He walked into an early group with John Collison, Patrick Collison, and Greg Brockman.

Stripe was finally going to talk to the credit card network more directly. On a partner call, the person in charge of certification said the next date was six weeks away and six weeks should be enough. In startup time, six weeks felt as far off as 10 years. Someone on the call asked whether any time was sooner. Tomorrow at 10 a.m. was open, along with a warning that the team was not going to be ready. They signed up anyway. The person on the call still assumed

Greg Brockman believes cybersecurity spend should vastly outstrip software spend. He says he may not be right, and he still holds the view. Software is becoming like the barrier to entry, and in the same exchange it is called cheap and close to free. People want software they can believe in, software that is reliable and secure. To him, resilience is the value, and it will matter even more.

The picture of one open-source maintainer as the little block under everyone else never made sense to Greg Brockman. Reliable software is hard to write, organizations depend on a supply chain, and few companies audit the supply chain. Uploading a malicious package has hit many companies. Human developers

Greg Brockman says OpenAI put about 25% of its production engineers on hold and told them every project was paused so they would only work on security.

OpenAI aimed its peak intelligence model at all of its infrastructure, built tools to repeat the check, found a number of issues, and fixed them. Brockman says the team saturated the issues that model could find, and a next model means they run the process again. Every time a new model drops, he says, it should be used to find vulnerabilities and then saturate them. OpenAI has also made a billion-dollar commitment so frontline defenders can access its models for defense. He says the internet was built for resilience more than for security, a

Greg Brockman says the defender's window is the current phase of cybersecurity, when peak frontier AI can help defenders secure their systems and the strongest capabilities are not broadly diffused yet. He says the physics of cybersecurity are changing. Cyber capabilities already available to everyone keep rising from model release to model release, from providers across the world.

The leading edges of impact are already showing up. In South Korea, high-profile hacks of banks were driven by different models. Brockman says defenders need to use the window while the strongest tools remain concentrated. Through the Daybreak program, he says his team has worked to give defenders responsible acc

Greg Brockman says debate and iterative amplification, techniques discussed since 2017 and two sides of a similar coin, apply when AIs solve problems too hard for humans to check. Summarizing a book is one example. It is hard to know whether a summary is any good unless the person has read the whole book.

If an AI only hands over a summary and calls it the summary, judging it is pretty hard unless another AI tries to prove the summary is bad. The second AI can point to a paragraph with a clearly important plot point the summary does not represent. A person does not have to read the whole book to see that the argument is strong. Brockman says ways of training AI can line up with human oversi

Greg Brockman says AI can resolve hundred year old math problems, and whether AI can come up with new knowledge now has a definitive yes.

He calls the math results a real milestone. Over the past thousand years, technology and scientific advancement have driven improvements in human quality of life, and he says the world now has the ability to massively accelerate scientific discovery. Models are not yet coming up with whole grand new theories. What excites him is what a human equipped with the tools can achieve that was unthinkable before. Human leverage to do whatever a person wants is going way up. If someone has a math problem to solve, that person can solve it. Early models had problem