Reports regarding the capabilities of the cybersecurity model are causing serious concern in the artificial intelligence model “Mythos” from Anthropic, which is said to have managed to infiltrate highly protected information systems within a few hours.
According to claims circulating in technology and security circles, the artificial intelligence system is said to have breached nearly all of the NSA’s classified information systems (National Security Agency of the U.S. ) within a matter of hours rather than weeks.
These reports are attributed to the head of the U.S. cybersecurity agency, who reportedly told The Economist that the model “had infiltrated nearly all of our classified systems, not in weeks, but in hours.”
These claims have sparked a debate about whether new artificial intelligence models could evolve into tools capable of automatically detecting vulnerabilities andexploit vulnerabilities even in highly secured networks.
This concern is amplified by the broader discussion surrounding AI security and the risk that advanced systems being used for large-scale cyberattacks.
What is Anthropic’s Mythos
Mythos is a new, very powerful artificial intelligence model from Anthropic, the company that created Claude AI. Unlike a simple chatbot, it is designed for more complex tasks, such as code analysis and troubleshooting computer systems.
The interest surrounding Mythos is primarily related to cybersecurity. This artificial intelligence can help experts find vulnerabilities in networks and programs before hackers exploit them.
However, this very capability raises concerns, as such a powerful tool could theoretically also be used for malicious attacks. That is why Anthropic is testing it with restricted access rather than as a simple public tool.
Simply put, Mythos is a “digital security expert” with immense capabilities, which can help protect systems but at the same time raises questions about how powerful artificial intelligence should be used.
