OpenAI announced its latest AI model, GPT-5.6 (codenamed Sol), with a significant twist: the federal government will determine which companies can access the technology. This marks a major expansion of Trump administration oversight of Silicon Valley AI development, creating an unusual gatekeeping arrangement between the company and regulators.

OpenAI positioned Sol as its most powerful model yet, specifically highlighting improvements in coding and cybersecurity tasks. However, the company expressed concern about long-term government control, stating that such oversight could prevent critical tools from reaching developers, enterprises, and security professionals who need them. Only government-approved companies will receive access, with no pathway for individual users.

The move follows similar restrictions placed on Anthropic, whose AI models received export controls after officials discovered unauthorized access by a South Korean telecom company with suspected Chinese ties. This represents a dramatic shift from Trump's campaign promises of deregulation, as his administration has become increasingly hands-on with AI governance.

Former Trump AI adviser Dean Ball criticized the rapid policy evolution, noting federal AI policy has swung from 'implausibly libertarian to increasingly draconian and opaque' in mere weeks. The arrangement essentially makes the government a bottleneck for cutting-edge AI deployment.