OpenAI has made a sweeping overhaul of its core operating principles for the first time since 2018. The key change is a shift away from its previous stance of putting artificial general intelligence (AGI) front and center, toward competitiveness, AI infrastructure and the broader technology ecosystem.

Business Insider reported on April 27 that OpenAI CEO Sam Altman (샘 알트먼) posted five principles for a frontier model research lab on a blog that day. The revision is seen as the biggest reset of standards since the 2018 charter.

The most notable shift is the reduced emphasis on AGI. The 2018 document mentioned AGI 12 times and aligned the organisation’s purpose and social responsibility around responding to AGI. In the new document, mentions of AGI drop to two. Instead, it stresses that society must respond at each level as AI capabilities spread in stages. OpenAI described this as an extension of its existing “iterative deployment strategy.”

Its stance on competition has also changed. In the past, it said it could stop competing and cooperate if another project came closer to safe AGI, but that language is absent from the new document. Instead, it stressed that as the company’s influence grows, it will disclose more transparently when it changes operating principles and why. It also said it would maintain its goal of broad-based prosperity, but could, if needed, reduce some empowerment and opt for greater resilience.

The shift also coincides with a recent competitive landscape. OpenAI has been in intense competition with Anthropic in recent months. Anthropic has raised its profile by advancing its “Claude” model, and there is an assessment that it has surpassed OpenAI in corporate valuation.

Another change in the document is the “audience” of its promises. While the 2018 charter emphasised principles and responsibilities within OpenAI, the new document is closer to recommendations aimed at the broader technology ecosystem. It stressed that AI-related decision-making should not be concentrated in a small number of labs and should be carried out more democratically, and urged governments to review new economic structures.

It also pointed to the need for large-scale global infrastructure to lower AI costs. This aligns with a recent intensification of competition to secure data centres and computing resources.

OpenAI has undergone changes since launching as a non-profit research organisation in December 2015, including the departure of some co-founders as it moved toward a for-profit structure. The revision is seen as resetting new operating standards suited to an expansion phase and market competition, moving beyond an early emphasis on the public interest and cooperation. Attention is focused on how these principles will be reflected in actual product launches, safety policies and responses to competitors.