OpenAI launched the AI Futures program on August 20, 2026. The research project, operated by the Strategic Futures Team, examines how societies can preserve individual rights and agency as artificial intelligence advances. The analysis of power concentration risks is at the center—according to the team, the most challenging long-term challenge in AI safety.
Power Concentration as Central Research Question
The Strategic Futures Team puts forward a fundamental thesis: Historically, political and military power relied on the cooperation of many people. Three factors were decisive—soldiers and police forces as bearers of the monopoly on violence, a civilian-military bureaucracy run by humans, and taxpayers who enabled state revenues through human labor and production.
The team points to David Hume's observation that those in power have "nothing to support them but [popular] opinion." Autonomous systems and machine intelligence could shift this balance: states could project power without relying on cooperative soldiers or police officers. Revenues would no longer come from human labor but from data centers. Bureaucracies could also be largely automated.
The team's conclusion: power could be maintained without a social contract agreed upon throughout society. The population could have little or no influence. "No amount of scientific discovery or technological progress or economic growth is worth the sacrifice of long-run individual autonomy this outcome would entail," the publication states.
ChatGPT Futures Class of 2026: First Generation Studying With AI
In parallel with the AI Futures program, OpenAI introduced the first ChatGPT Futures Class before June 9, 2026. Twenty-six students from over 20 institutions—including Vanderbilt, University of Toronto, Oxford, and Georgia Tech—each received a $10,000 scholarship plus access to OpenAI's frontier models.
The Class of 2026 began their studies in fall 2022, coinciding with ChatGPT's emergence. During campus visits, OpenAI observed that students used AI not to avoid work but to realize projects that previously seemed impossible. Examples include developing learning tools for classmates, translating mental health resources for underserved communities, scientific research, accessibility tools, and implementing side projects in organizations with measurable impact.
Kyle Scenna, 24, honoree and entrepreneur from University of Waterloo, is quoted: "I never thought the gap between noticing a problem and building something real could get this small." The program identifies three characteristics of this generation: they do not need to wait for expert status, funding, or permission before they start, build, or contribute.
GPT-5.5 and the Acceleration of Research
OpenAI released GPT-5.5 for Codex and ChatGPT before August 21, 2026. The model extends the acceleration in software development already observed last year to scientific research and general computer work. According to Artificial Analysis' Coding Index, GPT-5.5 achieves state-of-the-art intelligence at half the cost of competing frontier coding models. The model produces higher-quality outputs with fewer tokens and fewer attempts.
Long-Term Perspective: Superintelligence Amid Constant Everyday Life
An OpenAI representative stated in earlier publications that building superintelligence in ten years is "almost certain." At the same time, daily life is expected to feel surprisingly constant—"daily life and the things we care most about will change very little." The future will provide new ways to lead a fulfilling life, and more people will experience such life experiences than today.
The AI Futures program should not be confused with the AI Futures Project, founded in 2025 by Daniel Kokotajlo, a former researcher in OpenAI's Governance Division who departed in 2024.