Understand this story
On November 18, 2026, OpenAI CEO Sam Altman stated in a video on Semafor's YouTube channel that while OpenAI's speed in developing new models using its own technology is indeed faster than when not doing so, it does not involve simply pressing a button to have models design the next generation and rapidly self-improve without human oversight. He noted that OpenAI could have leaned more heavily into the recursive self-improvement (RSI) lineage but believed it was currently unsafe to do so. On the same day, OpenAI co-founder John Schulman argued in an A16Z podcast that AI alignment should not be controlled by a few companies but should become a global public science. OpenAI Chief Research Officer Mark Chen pointed out that the recursive self-improvement process is approximately 25% to 50%, AGI capability development is about 80%, and models still lack the ability to independently discover research directions.
AI-generated from public reports
Latest development · Report
OpenAI Co-founder on AI Decentralization: Alignment Should Become a Public ScienceDevelopments
2 confirmed developments
Sam Altman discusses OpenAI using RSI to build next-generation models
Source:Rohan PaulJohn Schulman argues that AI alignment should become a public science
Source:Wallstreetcn
Reports and discussion timeline
Browse reports and discussion by time. Their evidence levels remain distinct.
Nov 18
Wall Street CN
OpenAI Co-founder on AI Decentralization: Alignment Should Become a Public Science
Thinking Machines co-founder and OpenAI co-founder John Schulman stated in an A16Z podcast that AI alignment should not be controlled by a few companies but should become a global public science. OpenAI Chief Research Officer Mark Chen noted that the recursive self-improvement (RSI) process is approximately 25% to 50%, AGI capability development is about 80%, and models still lack the ability to independently discover research directions.
36Kr · AI
Paul Christiano warns of the risk of superintelligence going out of control, calls for pausing pretraining and reinforcement learning advances
In a podcast conversation with Dwarkesh Patel, Paul Christiano, a member of the OpenAI board's Safety and Security Committee and the inventor of RLHF, judged that on the current R&D trajectory, there is a one-in-three probability that within the next 12 months a superintelligence capable of overthrowing human rule will emerge, and that the risk of humanity permanently losing control is approaching 10%. He cited signals such as recent models producing hundreds of mathematical research results in a single month and machine learning progress lagging about a year behind pure mathematics, and reasoned that the acceleration in R&D could amount to jumping directly to the GPT-9 or GPT-10 level. He called on major frontier labs to pause algorithmic advances in pretraining and reinforcement learning, and to first spend 6 to 12 months rewriting safety software and locking down model weights and core secrets, and said that no single control scheme currently has better than a 50% chance of success.
Rohan Paul
Sam Altman on How OpenAI Uses RSI to Build Next-Generation Models
Sam Altman stated that OpenAI does indeed develop new models faster with the help of its own technology than without it, but it is not the case that with the push of a button a model designs the next generation under no human supervision and rapidly self-improves. He said OpenAI could have leaned further toward the RSI end of the spectrum, but believes it cannot currently do so safely. The relevant remarks come from a video on Semafor's YouTube channel.
Heat trend for this story
- Current
- 12
- Period peak
- 17 Nov 18 19:00
- 24-hour change
- —
Move the pointer or select the chart to inspect hourly heat. Use the left and right arrow keys with a keyboard.
AI Trends