In Major Security Shift, CISA Employs Anthropic’s Mythos AI to Audit Government Code
AI will soon become an enhancement to the security of federal digital systems. The Cybersecurity and Infrastructure Security Agency is currently using Mythos, an AI model created by Anthropic, to conduct searches within the federal software infrastructure looking for potential flaws.
The project, implemented by the Attack Surface Evaluation team at CISA, is focused on identifying any security blind spots before they can be used by foreign intelligence services or cyber criminals.
The new application marks a very practical step towards automation in national defense. In no way does the new technology replace human experts; rather, it serves as a fast code auditor. There have been reports that the AI system has discovered a great deal of flaws within code libraries.
The adoption of Mythos by the government, among other defensive tools, is even more impressive considering the turbulent political history of relations between Washington and the AI startup. Disputes arose when Anthropic first decided not to allow its models to be used for military and domestic surveillance activities, and as a result was called a supply chain risk by the Pentagon.
Even though the federal court subsequently overturned the label, it is worth noting that relations between the parties remained complicated and became even more difficult after an aggressive White House order, which briefly led to the suspension of the public version of the cyber-tool because of concerns of potential espionage activities.
However, federal institutions cannot help but notice the power of the innovation, and as a result, the National Security Agency is conducting tests with this software in a classified environment.
Relying on advanced algorithms capable of auditing government scripts, defense specialists can realize that they should utilize the technology driving the future to fight modern cyber threats. Thus, the application of automatic audits demonstrates an important change in the strategy of modern federal authorities.
