U.S. Government Reportedly Approves OpenAI’s GPT-5.6 Sol, Terra, and Luna Models

The Trump administration has reportedly lifted previous restrictions on the launch of OpenAI’s GPT-5.6, enabling a broader commercial rollout of this advanced model after weeks of government-mandated cybersecurity and national security vetting.

This decision marks a significant turning point in U.S. AI governance, moving GPT-5.6 from a tightly controlled pilot program with Trump-approved partners to a tool that will soon be available to a wider range of enterprises, developers, and security teams.

U.S. Approves OpenAI’s GPT-5.6 Models

Initially, the White House’s Office of the National Cyber Director and the Office of Science and Technology Policy urged OpenAI to limit access to GPT-5.6 to a small group of government-approved U.S. partners.

They invoked Executive Order 14409, which established a review framework for frontier models to assess the system’s cyber risk before any public release. Under this regime, OpenAI’s flagship model, GPT-5.6, along with its companion variants Terra and Luna, was restricted to a customer-by-customer vetting process.

Access was confined to roughly twenty entities, whose participation was officially disclosed to the federal government. This unprecedented pre-launch limitation was driven by concerns that highly capable multimodal models could enhance offensive operations, including automated vulnerability discovery, malware generation, deepfake-assisted social engineering, and large-scale disinformation campaigns targeting critical infrastructure and electoral processes.

According to reports from Axios, the Trump administration is now easing these constraints after its cybersecurity and technology offices completed an initial security and abuse-testing cycle on GPT-5.6 and its related tools.

Rather than a blanket release, the updated approach reportedly transitions GPT-5.6 into a structured rollout with enhanced guardrails: it will be more widely available for commercial and research users, but under stricter access policies, improved content filters, and mandatory logging for high-risk capabilities such as code synthesis and model-assisted red teaming.

OpenAI has indicated that it expects broader availability “in the coming weeks,” positioning GPT-5.6 deployment as a key test case for integrating frontier AI models into the economy while remaining subject to targeted federal oversight of cyber impacts.

For the cybersecurity community, the lifting of restrictions presents a double-edged sword. On one hand, defenders, Security Operations Center (SOC) teams, and threat intelligence groups will gain wider access to GPT-5.6’s advanced reasoning, multimodal analysis, and large-scale pattern-matching capabilities.

These features can be applied to tasks such as automated log triage, exploit chain reconstruction, phishing campaign clustering, and malware family classification.

On the other hand, the same capabilities may empower adversaries, lowering the barrier to entry for sophisticated social engineering, rapid exploit weaponization, and AI-assisted discovery of misconfigurations and zero-day vulnerabilities in complex cloud and operational technology (OT) environments.

With the Trump administration now signaling a willingness to relax controls once a frontier model clears an initial security review, GPT-5.6 is poised to play a central role in debates about the extent to which U.S. policymakers should regulate dual-use AI systems that can both enhance and threaten global cyber defense.

Interact with Cyber Threats in Windows, Linux, macOS VMs to Trigger Full Attack Chain - Analyse Malware & Phishing with ANY RUN