OpenAI has published a revised version of its “Our Principles” document, marking a significant shift in the company’s strategic priorities since its early days as a non-profit research lab. The new framework, released on Sunday, downplays the once-central goal of artificial general intelligence (AGI) and instead champions a wider, more rapid rollout of its technology.
In 2018, OpenAI’s mission was laser-focused on building AGI—AI that surpasses human intelligence—safely and for the benefit of all humanity. The original principles stated a “primary fiduciary duty is to humanity” and pledged to “minimise conflicts of interest” that could compromise broad benefit. The 2026 version, however, frames the challenge differently: society must now “contend with each successive level of AI capability, understand it, integrate it, and figure out the best path forward together.”
From Safety Caution to Democratic Access
CEO Sam Altman has been telegraphing this pivot for weeks. In a personal blog post earlier this month, he argued that AGI has a “ring of power” that “makes people do crazy things,” and that the only antidote is to “orient towards sharing the technology with people broadly, and for no one to have the ring.” The new principles reflect that philosophy, calling for democratising AI at all levels and resisting the concentration of power “in the hands of the few.”
One of the most striking changes is the removal of a 2018 commitment to step aside and assist any “value-aligned, safety-conscious project” that came closer to building AGI. The earlier document even specified a trigger: “a better-than-even chance of success in the next two years.” In 2026, there is no such promise. Instead, OpenAI acknowledges it is “a much larger force in the world than it was a few years ago” and pledges transparency about how its operating principles may evolve.
The shift comes amid intensifying competition. Rival Anthropic, valued at $800 billion (€696 billion) this month, refused to give the Trump administration unfettered access to its AI for military use in February, leading to a federal ban on Claude. OpenAI stepped in on February 28, signing a deal with the US Department of War—a move that prompted some users to boycott ChatGPT in favour of Claude. The episode underscores how OpenAI’s new principles may be shaped as much by market dynamics as by philosophical evolution.
The 2026 document also calls for broader societal changes to accommodate AI. It envisions “widespread flourishing at a level that is currently difficult to imagine,” but warns that this future depends on whether superintelligence is “held by a small handful of companies” or “in a decentralised way by people.” It reiterates policy suggestions such as “new economic models” and investments to drive down AI infrastructure costs.
OpenAI says it expects to work with governments, international agencies, and other AGI initiatives to “sufficiently solve serious alignment, safety or societal problems before proceeding further.” Practical examples include using ChatGPT to counter models that could create new pathogens or integrating cyber-resilient AI into critical infrastructure. The company has also launched GPT-Rosalind, a model tailored for European biotech research, signalling its intent to embed AI in specific continental ecosystems.
For European policymakers and businesses, the revised principles carry implications. The EU’s AI Act, which came into force in 2024, imposes strict requirements on high-risk AI systems, and OpenAI’s push for rapid deployment may clash with Brussels’ precautionary approach. The company’s call for “new economic models” echoes debates in Berlin and Paris about AI-driven labour market disruption, while its emphasis on decentralised control resonates with European concerns about digital sovereignty.
As OpenAI pivots from AGI obsession to practical rollout, the question for Europe is whether this new openness will translate into genuine partnership—or simply a more aggressive market entry. The answer may shape not just the continent’s AI landscape, but its broader technological autonomy.
