On April 15, foreign media outlets reported that OpenAI has granted select users access to trial its latest artificial intelligence model, GPT-5.4-Cyber. This model is specifically engineered to pinpoint software flaws and aid organizations in rectifying them. Just a week prior, its rival Anthropic introduced an AI tool called Mythos, which is dedicated to detecting and exploiting vulnerabilities within operating systems and web browsers. This move has raised apprehensions among financial institutions and governmental bodies. During a gathering of Wall Street leaders, organized by Bessent and Powell, executives were cautioned to treat Mythos with utmost seriousness. OpenAI intends to broaden the scope of participants in this early-access initiative, initially permitting several hundred users to evaluate the new model, with ambitions to extend the trial to thousands in the forthcoming weeks.