Trump Targets GPT-5.6 with Push for Tougher AI Policies

OpenAI has officially introduced its latest artificial intelligence family, GPT-5.6, but its rollout comes with an unusual condition. Before reaching the wider market, the three new models—Sol, Terra, and Luna—are being released under strict US government oversight due to national security considerations. Initially, only a carefully selected group of about 20 organizations, approved by federal authorities, can access the technology, with broader availability expected in phases. The flagship Sol model delivers the highest level of performance, Terra focuses on balancing capability with efficiency, while Luna is optimized for speed and lower operating costs. OpenAI has also unveiled an advanced “Ultra” feature that distributes complex tasks across multiple AI agents for improved performance. Despite complying with the temporary restrictions, the company emphasized that such government review should not become the permanent standard, warning it could slow innovation and limit access for developers, businesses, and cybersecurity professionals. OpenAI further stated that GPT-5.6 has been designed primarily to strengthen cyber defense by identifying and resolving software vulnerabilities rather than enabling offensive attacks. The limited release aligns with a new US policy framework that will evaluate advanced AI systems before they are widely deployed, marking a significant shift in how frontier AI technologies are regulated.