AI has reached a new cybersecurity milestone and this is what it means for your business
In a significant development for cybersecurity, Anthropic has announced Project Glasswing. Glasswing is an industry-wide initiative designed to secure critical software using next-generation AI capabilities.
At the centre of this initiative is a new frontier model, Claude Mythos Preview, which has demonstrated an unprecedented ability to identify and exploit software vulnerabilities and is reportedly outperforming all but the top echelon of highly skilled human security experts.
This marks a turning point in cybersecurity.
According to Anthropic, the Mythos model has reached a level where it can:
- Identify thousands of high-severity vulnerabilities across systems
- Detect flaws that have gone unnoticed for decades
- Develop increasingly sophisticated exploit paths
In testing, the model identified vulnerabilities across every major operating system and web browser, highlighting just how widespread and deeply embedded these risks are.
Cybersecurity has always been a race between attackers and defenders, however AI is poised to accelerate both sides.
Cybercrime already costs the global economy hundreds of billions of dollars annually, and AI-driven attacks stand to increase the frequency of attacks, reduce the time between discovery and exploitation, and enable lower-skilled threat actors to launch more sophisticated attacks.
For organisations, this is not a theoretical future risk, it is an immediate operational concern. Anthropic has not released this model to the general public, however given the rate of progression seen thus far in AI development it is foolish to consider these models being used nefariously as anything but just around the corner.
As tools similar to this become more widely available, businesses relying on legacy systems, running unpatched software, and lacking robust security processes will become increasingly exposed.
On the defensive side, Project Glasswing brings together major technology and cybersecurity organisations to apply these AI capabilities for the purpose of identifying vulnerabilities before attackers do and patch these critical systems across the industry.
If you are unsure whether your business is set up for success in this new AI-driven threat landscape, contact us to discuss how we can help you strengthen your security posture and stay ahead of emerging risks.