OpenAI limits latest ChatGPT product to Trump-approved customers (www.bnnbloomberg.ca)

OpenAI has announced a restricted release of its latest AI model, GPT-5.6 Sol, following a request from the Trump administration due to potential cybersecurity risks. The model is currently available only to a limited group of approved partners, a move criticized as an unprecedented level of government oversight in AI deployment. OpenAI expressed concern that such government vetting should not become the norm, framing this testing phase as a temporary measure before broader access is granted in the coming weeks. This decision comes in the wake of similar actions against OpenAI's competitor, Anthropic, whose new models were taken offline after government intervention due to security concerns about AI capabilities. This significant move underscores the increasing scrutiny by the U.S. government towards powerful AI tools, especially after warnings about their potential misuse in cybersecurity contexts. Trump’s recent executive order on AI oversight mandates a review of advanced AI systems for national security risks before public release. Critics in the tech community warn that the lack of a transparent process could hinder U.S. innovation and competitiveness in AI. OpenAI acknowledges that while its new model is designed to assist in identifying vulnerabilities, combining it with other tools could present unforeseen risks, highlighting the ongoing debate around the balance of innovation and safety in the rapidly evolving AI landscape.