Sam Altman unveils new AGI development principles as OpenAI shifts focus from superintelligence

Sam Altman Releases Updated Mission Statement for AGI Development

Sam Altman has published a new document titled "Our Principles" that outlines OpenAI's approach to Artificial General Intelligence and broader AI deployment 1

1

. The document, deliberately credited to Altman personally, presents five core principles: democratization, empowerment, universal prosperity, resilience, and adaptability 3

3

. "Power in the future can either be held by a small handful of companies using and controlling superintelligence, or it can be held in a decentralized way by people," Altman stated, emphasizing OpenAI's commitment to decentralizing AI power 3

3

.

OpenAI Shifts Away From Original AGI-Focused Mission

The most striking change in OpenAI's Our Principles document is the de-emphasis on achieving AGI, which was the company's founding purpose almost a decade ago 2

2

. While the 2018 version explicitly stated that OpenAI's "primary fiduciary duty is to humanity" and focused on building AGI safely and beneficially, the 2026 version treats AGI as just part of the company's wider AI rollout 2

2

. Just eleven months ago, Altman wrote on his personal blog that "the takeoff has started" and humanity was "close to building digital superintelligence" 1

1

. The language has softened significantly, suggesting the AGI rocket remains on the landing pad despite earlier predictions.

Democratisation of AI Access and Societal Adaptation to AI

Altman's democratization principle emphasizes that access alone isn't enough, arguing that decisions about AI development should be guided by democratic processes rather than controlled solely by major AI labs 3

3

. The document envisions "a world with widespread flourishing at a level that is currently difficult to imagine," where "a lot of the things we've only let ourselves dream about in sci-fi could become reality" 2

2

. To achieve universal prosperity, Altman noted that governments may need "new economic models" and significant investment to lower AI infrastructure costs 3

3

. OpenAI's strategy of "buying huge amounts of compute while our revenue is relatively small" and "vertically integrating to lower costs" reflects this commitment to making AI accessible 3

3

.

AI Safety Initiatives Face Scrutiny Amid Pentagon Agreement

The resilience principle addresses the benefits and risks of AI, acknowledging that "no AI lab can ensure a good future alone" . Altman specifically mentioned risks from "extremely capable models that make it easier to create a new pathogen," requiring society-wide defense approaches 1

1

. However, critics point to contradictions between stated principles and actions. Altman had previously pledged to allocate 20% of OpenAI's compute to a superalignment team focused on mitigating AI risk, but the team reportedly received only a fraction with outdated hardware as resources prioritized commercial products 4

4

. OpenAI's February 2026 Pentagon agreement raised additional concerns, as it didn't make the company's prohibitions on domestic mass surveillance and fully autonomous weapons legally binding 4

4

.

Trust Deficit and Power Consolidation Concerns

Notably absent from the 2026 document is OpenAI's 2018 commitment to step aside and assist any "value-aligned, safety-conscious project" that comes closer to building AGI 2

2

. Instead, the document acknowledges OpenAI "is a much larger force in the world than it was a few years ago" 2

2

. Following the November 2023 board crisis, Altman returned to a reshaped board more tightly aligned with him, and after advocacy groups questioned OpenAI's restructure, the company served legal notices to at least seven of them 4

4

. The adaptability principle states "we will learn quickly and course-correct," which suggests shipping, learning, and iterating rapidly 1

1

. This sits uneasily next to the document's heavy safety framing, as AI regulation experts note that safety implies restraint while scale implies speed. As one analysis concluded, there's clearly a trust deficit between Altman and the industry, and earning back that trust will require actions that speak louder than words posted on ChatGPT maker's website 4

4

.