U.S. President Donald Trump talks about the risk of ​artificial intelligence (AI) to the banking system and acknowledges the requirement for government safeguards. However, Trump also talked about the benefits of AI and how it could make banking operations safer and more secure. 

Fox Business Network (via Reuters) in an interview asked Trump about whether AI could be trusted for the banking system, to which he said: “Yeah, ​probably.” 

But he also added, “It could also be the kind of technology that allows greatness in the banking ​system, makes it better and safer and ​more secure.”

Trump also supported the idea that governments should have a way to control or shut down AI if needed to prevent risks or misuse. In addition, concerns also persist over whether the technology could end up undermining the very systems it is meant to secure.

The comment came after growing concerns about Anthropic’s latest cybersecurity model, “Claude Mythos.” The company claims the model has the ability to detect cyberthreats, flaws, and vulnerabilities in critical software systems. However, media reports suggest it also raises concerns about the misuse and how it could challenge the banking industry. 

Due to the advanced capabilities of Mythos, Anthropic has released the preview to limited companies under Project Glasswing. It states that the model needs to be tested and studied if it plans to make the model available to the public. 

Anthropic is reportedly in talks with the Trump administration about Mythos and its broader frontier‑AI roadmap, with a co‑founder saying the company believes the U.S. government must understand such advanced models and will continue to engage on future ones as well.

Just a week after Claude Mythos debut, its rival OpenAI also launched GPT 5.4 Cyber model, a tailored version of GPT 5.4 to strengthen AI-driven cybersecurity. The model is designed for security professionals to detect vulnerabilities, analyse threats, and build more advanced defensive systems.