OpenAI is entering its Phase Three, shifting from building advanced AI to making it affordable, accessible, and useful for everyone, with the aim of delivering personal AGI to every human.

The plan centers on AI-assisted research, including automated AI researchers that could collaborate with humans and potentially handle a large share of research by 2028.

Three long-term goals drive the initiative: automated AI researcher, accelerated scientific and economic progress, and universal access to a personal AGI assistant.

CEO Sam Altman and chief researcher Jakub Pachocki caution that fully automating all tasks would be dangerous and undesirable; humans will decide what's worth doing.

OpenAI confidentially filed for a US IPO with the SEC, signaling readiness for broader governance and public markets, alongside industry peers.

The timing aligns with a confidential S-1 filing for a potential IPO, accompanying the blog post and indicating preparation without a firm timetable.

The company emphasizes progress toward public markets through confidential IPO paperwork, underscoring governance considerations while keeping an IPO date undecided.

Humans remain central: AI should assist and augment human goals, with people setting purposes, trade-offs, and deciding what is worth pursuing.

OpenAI argues for broad distribution of AI-generated prosperity and warns against concentrating AI power in a few institutions, aiming for a diverse ecosystem of builders and beneficiaries.

AI should tackle real-world problems—such as medical costs, lifelong learning, entrepreneurship, elder care, and financial decisions—rather than exist merely as a milestone.

An international body is proposed to coordinate and potentially slow frontier AI development to manage safety, resilience, and alignment risks, with emphasis on global cooperation.

The plan stresses safety and alignment, with humans overseeing powerful AI systems and AI augmenting rather than replacing human judgment, coupled with a push for global coordination to reduce catastrophic risk.