OpenAI Limits GPT-5.6 Sol Access During US Security Review

OpenAI has delayed the public release of its newest artificial intelligence model, GPT-5.6 Sol, after agreeing to a cybersecurity review requested by the Trump administration.

The company announced that the model will initially be available only to a small group of government-approved partners while federal officials evaluate its potential national security risks.

OpenAI adopts phased release for GPT-5.6 Sol

According to OpenAI, GPT-5.6 Sol will launch through a limited rollout while additional testing is completed.

The company said it does not believe government vetting should become the standard process for releasing AI products. However, it described the current review as a temporary measure before expanding access in the coming weeks.

OpenAI also stated that the new model offers stronger capabilities for identifying and fixing cybersecurity vulnerabilities than for conducting cyberattacks.

Trump administration increases oversight of advanced AI

The limited launch follows an executive order signed earlier this month by President Donald Trump establishing a framework for reviewing the national security implications of advanced artificial intelligence systems before public deployment.

Although participation is officially voluntary, the order allows federal agencies to evaluate frontier AI models for up to 30 days before wider release.

The administration has argued that increasingly powerful AI systems could pose cybersecurity threats if malicious actors exploit their capabilities.

Anthropic faced similar government restrictions

OpenAI is not the first AI developer affected by the new oversight process.

Earlier this month, Anthropic temporarily withdrew two of its latest AI models, Fable 5 and Mythos 5, after receiving a directive from the Trump administration restricting their availability to foreign users.

Government concern intensified after Anthropic warned that one of its models demonstrated advanced capabilities for identifying software vulnerabilities that could potentially be weaponized.

OpenAI says safeguards accompany stronger AI

Despite the government review, OpenAI maintains that GPT-5.6 Sol does not exceed the company’s internal safety thresholds.

The company acknowledged, however, that uncertainty remains whenever increasingly capable AI systems interact with other technologies.

For that reason, OpenAI said it paired the model’s improved performance with additional safeguards and a gradual release strategy designed to monitor real-world behavior before expanding access.

So far, the company has not disclosed the identities of the roughly 20 approved customers testing the system.

Experts question government intervention

Some cybersecurity specialists argue that increased government involvement could slow American AI innovation.

Several researchers have criticized recent actions against Anthropic, saying there is little evidence that the restricted models present risks beyond those already associated with publicly available AI systems.

Critics also warn that limiting access to cutting-edge AI products could weaken the competitiveness of US companies as international rivals, including Chinese developers, continue advancing their own technologies.

Negotiations continue between AI companies and Washington

OpenAI CEO Sam Altman recently met with US Commerce Secretary Howard Lutnick as discussions continue between federal officials and leading AI companies over future oversight rules.

Anthropic has also participated in negotiations, although its relationship with the administration has become increasingly contentious after disputes over AI safety policies and government contracts.

Federal agencies continue developing the long-term framework that will determine how advanced AI models are reviewed before public release.

AI regulation enters a new phase

The limited release of GPT-5.6 Sol highlights the growing tension between accelerating AI innovation and addressing national security concerns.

As governments seek greater oversight of increasingly powerful AI systems, technology companies face the challenge of balancing rapid product development with evolving regulatory expectations.