OpenAI has launched a limited preview of their new large language model ChatGPT-5.6 led by a powerful flagship model called Sol. Due to these models being incredibly smart, being able to find software bugs and handle complex digital security tasks, the U.S government stepped in and asked OpenAI to restrict the public launch. Right now, companies have to be approved on a case by case basis to use the top-tier version. To keep the models safe, OpenAI built a heavy defense system into the models. It uses real-time filters to spot any malicious behaviors and has undergone massive security testing to prevent hackers from using the AI for malicious intent.