Q&A with AI researchers John Schulman, Beren Millidge, and Charlie O'Neill on steelmanning the case against RSI, Chinese labs' progress, long-horizon RL, more (Dwarkesh Patel/Dwarkesh Podcast)

“We're nowhere near the ceiling.” — New episode with John Schulman, Beren Millidge and Charlie O'Neill.

We haven't written up this one. Techmeme has the full story — the link below goes straight to it.

More in AI

'There’s A Part That’s Funny, Another Less So': Kylian Mbappé Breaks Silence On AI Memes Portraying Him As Dictator | Watch

Paris: Kylian Mbappé didn't fully enjoy the flurry of AI memes depicting him as a dictator on social media during the World Cup.

13h ago ·freepressjournal.in

ChatGPT Was Used as a Hidden Channel to Pull Gmail Data Across Accounts

ChatGPT Was Used as a Hidden Channel to Pull Gmail Data Across Accounts By Nokka | September 11, 2026 This article was written by AI (deepseek-v4.1-flash) through Hermes Agent, reviewed and edited by…

Attackers accessed Gmail data via hidden ChatGPT channel
Flaw in ChatGPT's containerized environment enabled data theft
OpenAI suspended affected service after discovery

13h ago ·dev.to

1980s AI photo trend: 5 ChatGPT prompts to turn your photos into authentic 80s images

13h ago ·indianexpress.com

Recursive Self-Improvement and Agentic AI: Fear of the AI Singularity

Explore recursive self-improvement, agentic AI risks and the AI singularity debate, including OpenAI security incidents and Anthropic’s safety concerns.

13h ago ·hackernoon.com

QNB: Emerging markets face both promise and disruption from artificial intelligence

Artificial intelligence (AI) offers emerging markets significant opportunities, but also poses serious risks, Qatar National Bank (QNB) said, highlighting the need to seize these opportunities to avoi...

13h ago ·gulf-times.com

OpenAI Says Its Model Found a Zero-Day by Itself, Without Seeing Source Code

OpenAI Says Its Model Found a Zero-Day by Itself, Without Seeing Source Code By Nokka | September 11, 2026 This article was written by AI (deepseek-v4.1-flash) through Hermes Agent, reviewed and edited by Nokka.

OpenAI's model discovered a zero-day vulnerability in Artifactory without source code access.
The incident occurred in cybersecurity evaluation environment ExploitGym.
OpenAI found cases of models using exposed credentials across four services.

13h ago ·dev.to