OpenAI limits GPT-5.6 Sol release at government request as U.S. reviews cybersecurity risks tied to the powerful new AI model

OpenAI's GPT-5.6 Sol model is facing restrictions as the U.S. government reviews its cybersecurity implications. The company has begun a limited rollout to a select group of partners, a decision made at the government's request, highlighting ongoing concerns about AI oversight.¹⁴⁶

The White House has asserted control over advanced AI models, with tech giants like OpenAI agreeing to governmental oversight. OpenAI CEO Sam Altman expressed concerns about the government selecting customers for AI tools, emphasizing national security.²⁵

A June executive order indicates that while advanced AI can bolster national strength, it necessitates coordination across federal agencies and private companies. Critics fear this could lead to government control over access to cutting-edge AI technologies.⁷

OpenAI's decision to limit access to GPT-5.6 Sol comes as U.S. officials assess potential cybersecurity risks. The model, which includes specialized sub-models like Terra and Luna, is designed to outperform competitors like Anthropic at lower costs. However, full public access hinges on regulatory approval.

Despite the limited preview for trusted partners, there are rumors that a broader launch could occur soon. Altman noted in an internal memo that complying with government reviews is not the company's preferred long-term strategy, aiming for a wider release in the coming weeks.

OpenAI claims that Sol is particularly adept at identifying and addressing software vulnerabilities, rather than executing full cyberattacks, which could be a crucial factor in the ongoing discussions about AI's role in cybersecurity.