📚 New to this topic? Read our full guide: What Is Artificial Intelligence?.
OpenAI’s GPT-5.6 Now Needs US Government Approval for Each Customer
OpenAI’s new GPT-5.6 model now requires US government approval before it can reach customers, one by one. The Trump administration has asked OpenAI to “stagger,” or slow down, the release of its powerful new AI model over safety worries. This is the second time the US has stepped in to control how a top AI model spreads around the world.
The plan is unusual. Instead of a normal wide launch, OpenAI will do a limited release first. A broader rollout will follow “a couple of weeks later.” During the early preview, the government will sign off on access on a customer-by-customer basis. In plain words, officials get a say over who can use the model first.
Why is the White House slowing the rollout?
The main worry is cybersecurity. Cybersecurity means protecting computers and networks from attacks. Officials fear that very advanced AI models can find and exploit software weaknesses faster than human experts. A “weakness” or vulnerability is a hidden flaw in software that hackers can use to break in.
These so-called frontier cyber models could, in theory, scan enterprise networks and find hidden entry points at high speed. That power could help defenders, but it could also help attackers. Reports point to Anthropic’s “Mythos” model, where independent tests confirmed advanced AI could “develop real browser exploits autonomously.” Autonomously means the AI did it on its own, without a human guiding each step.
How the approval process works
OpenAI CEO Sam Altman reportedly told staff that during the preview, the government would be “approving access customer by customer.” OpenAI staffers worked closely with officials on the release plan. Two government bodies are involved: the Office of the National Cyber Director and the Office of Science and Technology Policy.
OpenAI made clear it does not love this setup for the long run. The company said: “this is not our preferred long term model, and will work with them and others in industry to achieve a more sustainable approach for future releases.” So the customer-by-customer approval is meant to be temporary.
Key facts
| Detail | Information |
|---|---|
| Model | OpenAI GPT-5.6 |
| Approval type | Customer-by-customer during preview |
| Main concern | Cybersecurity / autonomous exploits |
| Release plan | Limited preview, broader release ~2 weeks later |
| Government bodies | Office of National Cyber Director; Office of Science and Technology Policy |
| OpenAI CEO | Sam Altman |
| Commerce Secretary cited | Howard Lutnick |
The “second attempt” to control the AI race
This is being called a second attempt by the US to steer the global AI race. The Trump administration earlier signed an executive order. An executive order is a direct rule from the President. That order asked AI companies to voluntarily submit new models for government testing before they go public. That was a shift from the government’s earlier hands-off approach.
Although the review is called “voluntary,” reports say it has real teeth. A White House spokesperson said the administration “continues to collaborate with frontier AI labs to develop shared approaches for addressing the challenges of scaling this technology.”
Why it matters (especially for India and founders)
If the US controls who gets early access to top models, that affects the whole world, including India. It could mean Indian companies wait longer to use the newest tools, or face extra checks. This is part of a bigger trend where governments are gatekeeping AI. We have already seen India in talks with the US to access Anthropic’s Fable 5 model, which shows access is now a diplomatic issue.
For founders, the message is clear. The best models may not always be freely available. Building a business fully dependent on one frontier model now carries political risk. The same cyber and safety fears even shaped how OpenAI’s rival Anthropic grew, as our piece on Claude winning paid consumers notes.
FAQ
Is the government approval permanent?
No. OpenAI said the customer-by-customer approval is not its preferred long-term model and wants a more sustainable approach for future releases.
What is a “frontier” AI model?
A frontier model is one of the most advanced AI systems available, pushing the limits of what AI can do. These models draw the most safety attention.
Why does cybersecurity drive this concern?
Advanced models can reportedly find and exploit software flaws on their own. That power could be misused to attack networks, which is why the government wants oversight.
The takeaway is that AI is no longer just a tech story. It is now a national security story. When a model as powerful as GPT-5.6 needs government sign-off for each customer, it shows how closely the US wants to manage the AI race. The rest of the world, including India, will be watching what this means for fair access.
Sources: Financial Express, The Decoder, and TechCrunch.
