AI War Heats Up as Two Giants Showcase New Technologies

The AI war is heating up after several companies launched their latest models. One of them is the technology giant OpenAI, which has just released GPT-5.4-Cyber. This latest variant works specifically for defence in cybersecurity, cited from Reuters on Wednesday (15/4/2026). The company has launched it on a limited basis at this initial stage due to its permissive design. Namely, for verified security vendors, organisations, and researchers. In addition, OpenAI is expanding the Trusted Access for Cyber (TAC or Trusted Access for Cybersecurity) programme. Now, thousands of individual defenders and hundreds of teams protecting critical software that have been verified can use it. There is a new tier added in this programme. Reuters notes that higher verification levels unlock more advanced capabilities. Verified users will gain access to the latest GPT-5.4-Cyber model. However, with usage restrictions on sensitive matters such as research and vulnerability analysis. The launch of OpenAI’s latest model coincides with Anthropic also releasing Mythos. This AI model, part of Project Glasswing, can also be used for defensive cybersecurity. However, this launch has encountered issues because Mythos is reported to be very powerful and capable of identifying unknown vulnerability gaps. Anthropic has decided to delay the launch to a wider user base. The company is also arranging meetings with US officials to discuss the cybersecurity capabilities created by Mythos. It is reported that Mythos can only be used by 40 technology companies so far, including tech giants such as Microsoft and Google.