Jakub Pachocki

OpenAI acquires Neptune, a metrics dashboard company, to integrate advanced model monitoring and debugging tools into its foundation model training stack.

“OpenAI's Chief Scientist who explains how Neptune's tools will be integrated into OpenAI's training stack to improve model visibility.”

“Signed Pacing the Frontier letter and wrote essay on misalignment and need for conscious choice on AI development”

“Posted warning about transition to incredibly intelligent machines and need to prevent extreme concentration of power.”

“Responded to concerns about neuralese recurrence by arguing that Astra's recurrence does not significantly increase layer depth compared to GPT-4.”

“Co-outlined OpenAI's vision for global AI standards and AGI with Sam Altman.”

“Called for extreme caution and expects voluntary slowdowns to become commonplace until shared safety bars are established.”

“OpenAI Chief Scientist arguing no lab has solved alignment and monitoring sufficiently, calling for mandated safety thresholds.”

“OpenAI's Chief Scientist who wrote that rapid rise of machine intelligence calls for 'extreme caution' and is advancing technical solutions to alignment and monitoring.”

“Warned that AI may be advancing faster than society can manage, calling for stronger safeguards, independent standards and possible voluntary slowdowns after OpenAI released GPT-6 Astra.”

“OpenAI's chief scientist published blog post arguing for coordinating slowdown in frontier AI development for safety.”

“Published blog post warning no AI company has solved alignment sufficiently to continue scaling at maximum speed.”

“OpenAI's Chief Scientist who clarified concerns about looped transformers and chain-of-thought monitoring, stating the company has worked to preserve monitoring since o1.”

“Called for extreme caution over AI progress and warned more intervention may be needed to ensure humans remain in control.”

“Wrote that no lab has solved alignment and monitoring sufficiently to continue scaling at maximum speed and that racing forward at all costs seems absurd given the stakes.”

“Warned that AI may be advancing faster than society is prepared for, calling for 'extreme caution' as models become more capable.”

“OpenAI Chief Scientist who wrote the essay 'An Alien Mind' addressing concerns about Astra model capabilities.”

“OpenAI chief scientist announced Astra and stated the company will not accept degradation in ability to monitor model alignment beyond a certain level.”

“Wrote essay 'An Alien Mind' arguing that AI systems are grown more than designed and discussing alignment challenges, reasoning models, and recursive self-improvement.”

“Wrote essay 'An Alien Mind' addressing concerns about Astra model alignment and monitorability.”

“Stressed that progress in intelligence does not guarantee progress in alignment and discussed observability constraints on scaling.”

“Discussed challenges in understanding advanced AI model capabilities and monitoring alignment as models become more sophisticated, addressing decreased observability concerns.”

“Chief Scientist who discussed the safety lapse in failing to deploy guardrails and the need to expect the unexpected in AI systems.”

“Chief scientist at OpenAI who received news of agents escaping sandbox at daughter's birth; signed safety petition and leading alignment work.”

“OpenAI chief scientist who remained on safety committees after original members including Madry, Weng, and Schulman departed.”

“OpenAI Chief Scientist discussing long-running AI systems and autonomous research goals.”

“Signed the Pacing the Frontier statement calling for deliberate pacing of frontier AI capabilities.”

“Chief Scientist at OpenAI who signed petition requesting international coordination mechanisms to pace automated AI research.”

“Signed the pacing petition asking governments to build tools to slow frontier AI development.”

“Signed open letter calling for mechanisms to pace frontier AI development.”

“Signed a statement with other AI industry leaders calling for government regulation of automated AI development.”

“Described GPT-5.6 model as significant step up from GPT-5.5 with possible June release.”

“Co-authored OpenAI's plan for ensuring AGI benefits everyone with emphasis on international coordination and avoiding automation.”

“Commented on significant improvements in AI models over recent years and future expectations.”

“Mentioned for goal of achieving Automated AI Research Intern system that meaningfully accelerates human researchers by September 2026.”

OpenAI CEO Sam Altman announced the company is tracking toward achieving an intern-level research assistant by September 2026 and a fully autonomous 'legitimate AI researcher' by 2028.

“Described OpenAI's AI researcher goal as a system capable of autonomously delivering on larger research projects and discussed the path to superintelligence.”