The White House has compelled OpenAI to restrict the release of its advanced GPT 5.6 model to a select group of government-approved partners, solidifying state control over critical technological infrastructure. This move follows a similar intervention by the administration against Anthropic, which saw its advanced Mythos and Fable models pulled from release under an export control order. The actions underscore a concerted effort by the state to manage the development and deployment of artificial intelligence, ensuring its alignment with existing power structures and the interests of concentrated capital, rather than allowing for broader, uncontrolled access.
OpenAI agreed to limit the model’s release, with CEO Sam Altman noting in a memo sent to the company on Thursday that the government is approving access “customer by customer.” Altman stated this is “not our preferred long term model,” indicating a tension between corporate desire for market expansion and state demands for control. The administration views OpenAI’s latest model as “on par” with Anthropic’s previously suppressed Mythos, highlighting the perceived threat these advanced capabilities pose to established order if not contained.
The Commerce Department’s earlier export control order on Anthropic’s Mythos and Fable models was prompted by fears in Washington and on Wall Street over their advanced cybersecurity capabilities. These capabilities, framed as potential “unprecedented safety risks,” effectively led to the suppression of technologies that could potentially disrupt existing surveillance and control mechanisms, or be utilized outside the purview of state-corporate oversight. The state's intervention demonstrates its primary function: to protect accumulated wealth and suppress challenges to the existing distribution of power, even when those challenges are technological advancements.
The State's Hand in Capital's Future
A White House official stated the administration continues “to collaborate with frontier AI labs to develop shared approaches for addressing the challenges of scaling this technology.” This collaboration, however, is not a neutral partnership but a mechanism to integrate private capital into the state's security apparatus, ensuring that technological progress serves the interests of the ruling class. The ad hoc nature of these interventions, occurring in a “strange moment” with no true federal regulatory framework, reveals the state’s capacity for swift, decisive action when perceived threats to its authority or the stability of capital arise.
Managing Contradictions, Preserving Power
President Donald Trump signed an executive order earlier this month asking AI companies with advanced models to voluntarily submit them for government review 30 days before release. This voluntary framework, however, remains largely unestablished, creating confusion among AI companies regarding the specific agency directing AI regulation. Despite this lack of a clear, consistent framework, the state has not hesitated to exert direct control, as seen with the White House’s request to OpenAI and the Commerce Department’s ban on Anthropic. This demonstrates that formal legal structures are secondary to the immediate needs of capital and state power.
Brad Carson, head of Public First, a bipartisan pro-AI safety super PAC, remarked last week that “The Fable episode shows the need for clear regulations.” Carson added, “Right now, you have an ad hoc, personalized, opaque, possibly lawless approach.” While calling for “transparency and basic fairness,” such reform efforts within the current system extend its life without addressing its foundations. The focus remains on managing the system’s contradictions through regulation, rather than questioning the fundamental concentration of power and control over critical technologies by a select few. The government's immediate recall of "dangerous products, including AI models," is presented as appropriate, yet the criteria for "dangerous" are defined by those in power, serving to protect their interests.
The current situation highlights how the state, through its various departments, acts as an enforcer for capital, ensuring that advanced technological capabilities are not allowed to develop in ways that might challenge the existing economic order. The suppression of certain AI models and the restricted release of others serve to maintain a controlled environment where innovation is permitted only insofar as it reinforces, rather than undermines, the concentration of wealth and power.