OpenAI has closed its AGI Readiness Team, a group focused on safeguards for advanced artificial intelligence systems. The move adds another major internal change to a company already facing questions about how it balances product development, safety work, leadership turnover and outside scrutiny.
What OpenAI changed
The AGI Readiness Team worked on the safety of artificial general intelligence, or AGI. OpenAI defines AGI in economic terms: AI systems that can operate autonomously and automate a wide range of human tasks.
According to the source, the team is not continuing as a standalone group. Its members will be reassigned to other departments within OpenAI.
That detail matters because AGI readiness is not a narrow engineering issue. It touches planning, oversight, deployment, governance and the question of whether powerful AI systems can be introduced without creating risks that institutions are not prepared to manage.
Brundage says the readiness gap is serious
Miles Brundage, OpenAI's outgoing Senior Advisor for AGI Readiness, announced that he is leaving the company and raised strong concerns about the state of preparation around advanced AI.
"In short, neither OpenAI nor any other frontier lab is ready, and the world is also not ready,"
Brundage’s warning is broad. It is not framed only as a criticism of one company or one team decision. His point, as reported in the source, is that readiness is missing across frontier labs and across society more generally.
He also points to gaps in AI oversight. In his view, technology companies have strong financial motivations to resist effective regulation. That creates a difficult alignment problem: the companies building advanced systems may not naturally move toward the level of external control or restraint that critics believe is needed.
Brundage argues that safe AI systems will not appear automatically. The source says he believes governments, companies and civil society all need to act deliberately.
"I think AI is unlikely to be as safe and beneficial as possible without a concerted effort to make it so."
After leaving OpenAI, Brundage plans to either establish or join a non-profit organization. He says he can have more impact from outside the industry. His team also developed the five stages of AI progress for OpenAI.
Another safety team closure follows the Superalignment move
The AGI Readiness Team shutdown follows OpenAI's decision in May to disband its Superalignment team. That group studied long-term AI safety risks.
At the time, team leader Jan Leike publicly criticized the company. He said that "security culture and processes have to take a back seat to "shiny products." The source presents the AGI Readiness closure as part of a wider pattern of safety-focused teams being dismantled or reorganized inside OpenAI.
For outside observers, that pattern raises a practical question: where does dedicated advanced AI safety work sit inside the company now? Reassigning staff does not necessarily mean the work disappears, but it changes the structure around it. A separate team can concentrate attention, define priorities and give readiness work an institutional identity. Moving people into other departments may spread expertise, but it can also make accountability harder to see from the outside.
Leadership churn and investor confidence are moving together
The company has also seen additional leadership changes. Three executives departed last month, including CTO Mira Murati and Head of Research Bob McGrew.
At the same time, investors recently value OpenAI at $157 billion, despite internal changes and ongoing financial losses. That combination is central to the broader debate around OpenAI: the company remains highly valued while critics question whether its governance and safety structures are keeping pace with its ambitions.
The source describes two possible conclusions critics draw from the shutdown of OpenAI's AGI and ASI security teams:
- OpenAI may no longer see advanced AI as an immediate possibility.
- OpenAI may be taking an irresponsible approach to AI safety.
Those interpretations point in very different directions. If advanced AI is not close, then reorganizing readiness work could look like a shift in priorities. If advanced AI is close, then reducing dedicated readiness structures could look far more consequential.
The timeline debate keeps pressure on safety planning
Sam Altman continues to argue that AGI is both feasible and imminent. That position keeps the safety debate urgent, because a company that sees AGI as near must also explain how it is preparing for the systems it says may arrive.
The source also notes that Dario Amodei, CEO of Anthropic, one of OpenAI's main competitors, suggested that AGI could emerge as early as 2026, while saying the timeline remains highly uncertain.
That uncertainty is part of the challenge. If AGI arrives later than expected, readiness work may look premature to some business leaders. If it arrives sooner, the absence of strong preparation could become a major failure. Brundage’s argument is that readiness requires intentional effort before the moment of deployment, not after the risks are already visible.
OpenAI’s decision to shut down the AGI Readiness Team therefore lands in a larger dispute over advanced AI governance. The facts in the source show a company reorganizing safety-related work, losing senior figures, retaining high investor confidence and continuing to speak about AGI as a serious near-term goal. The unresolved question is whether the structures around that goal are becoming stronger, weaker or simply harder to evaluate from the outside.