A New Era in the AI Race: US Government Regulates GPT-5.6 and Other Models

The competition between two giants in the world of artificial intelligence — OpenAI and Anthropic — has now entered a completely new stage. It is no longer just about technological superiority, but about survival under state control. The US government has begun to establish strict control over the release of the most advanced AI models to the public, which could jeopardize the future of the entire industry. This is reported by Techcrunch.com reports it says.

Recently, Anthropic's Fable and Mythos models were halted by the government. Now, OpenAI's anticipated new GPT-5.6 model is expected to face a similar fate. According to The Information, this model will be available only to a limited circle of clients, and separate government permission will be required for each user. While Sam Altman estimates this process will take a few weeks, the Anthropic experience suggests this period could stretch into months.

Control and Economic Risks

Such government intervention could be a serious economic blow to AI laboratories. Delaying the market entry of new systems into which billions of dollars have been invested reduces the profitability of companies. This, in turn, negatively affects the construction of data processing centers and the development of technological infrastructure. If the process continues in this manner, it is not surprising that stagnation could occur across the entire sector.

Currently, OpenAI and Anthropic have fallen into the same problematic situation. Various accusations are echoing among industry representatives: some accuse Anthropic of stifling competitors through regulations, while others suspect OpenAI leadership of getting close to political circles to squeeze rivals out of the market. However, the situation has become far more serious than personal competition.

Security Issues and Future Steps

Although the US government intends to inspect the models, regulatory bodies currently lack sufficient experience and technical capabilities to test such complex systems. Most importantly, clear criteria have not been developed regarding exactly which risks the government wants to protect against. Nevertheless, the capabilities of AI in terms of cybersecurity and biological risks are fueling state-level concerns.

Industry experts, including Dean Ball, who is expected to join the OpenAI team soon, believe that industry representatives should unite in the following directions:

• Trusting independent expert groups to manage the oversight processes;
• Supporting the most optimal regulatory options instead of fighting every new rule;
• Protecting AI not as the interest of individual companies, but as an entire industry;
• Not turning security issues into a weapon to strike competitors.

AI models have reached a level where their capabilities now have direct political and international consequences. To mitigate these consequences and continue technological progress, industry representatives are required to set aside competition and act collectively. In the coming weeks, it will become clear whether the AI industry is capable of such cooperation.