Updates from C7: securing the AI landscape

Over the last few weeks, the Container 7 team has continued to track emerging threats and technologies, uncovering how adversaries are weaponizing AI and other technologies across the wider threat landscape. The team has been identifying and exposing vulnerabilities and other key concerns in legacy infrastructure and AI systems, with the aim of supporting our customers in identifying, preventing, and mitigating threats (from humans or AI) before they materialize.

The last few weeks have seen a number of vulnerabilities, exploits, and compromises dominating news cycles, and the team has been staying on top of them. Alongside releasing a number of new pieces of content to help organizations bolster their cyber and AI readiness, some of the areas C7 has been tracking include:

• The PHP2Shell vulnerabilities (a pair of critical flaws in PHP, rated 9.2 on the CVSS v4.0 scale) where an unauthenticated attacker can utilize these vulnerabilities to run code on the server, and your application’s own login can’t stop it; the impact of this has been huge as it potentially affects around 70% of websites. The most concerning part is that for many teams, patching alone may not fully harden organizations to the potential of exploitation, as certain versions are at the end of service life. As always the researchers who found these bugs used AI models and attackers have the same tools.
• The Iris/C2Agent, a new ‘goal-driven’ Malware where a threat actor can state a target such as ‘gain administrator password’ and, rather than a threat actor guiding the command and control via a series of prompts, the AI will work away on achieving the objective, with newer variants even hijacking AI developer tools already on the machine and using legitimate commercial AI services to do the work.
• The Binance compromise, the largest-ever crypto breach, with around $6.3 billion in assets stolen, was also identified last week. 

It is safe to say that new vulnerabilities are being identified and exploited at an increasing pace, and keeping up with them all and how they potentially impact our customers drives our team each day. As always, we continue to emphasize the importance of strong cyber resilience by maintaining and evolving robust vulnerability management processes and asset inventories, ensuring RBAC and least privilege, embedding the secure-by-design principle, and threat modeling the challenges you are likely to face.

Our belief is that a strong security posture is built through organizational cyber resilience, whether against threats organizations have faced for years or the nuances of the growing number of AI-associated threats.

We continue to see organizations balancing the challenges of enabling business optimization by connecting AI agents to production systems and building pipelines, while trying to ensure they are effective at assuring, governing, and tracking risks associated with them. The trade-offs between driving business growth and ensuring you do it securely continue to be competing priorities, and our mission is to support our customers through our extensive catalog aimed across the workforce, from first line to the executives.

Want to hear more? Reach out to sales@immersivelabs.com to learn more about our releases from the last few weeks and other exciting updates and insight, including: 

• PHP2Shell, Iris, or other key focus areas
• Hands-on content aimed at the whole workforce – released both in a vendor-agnostic capacity and in a Claude code variant
• The ROI and use cases for optimizing through AI
• The risk of Shadow AI
• The complexities of copyright when utilizing AI functionality 
• The nuances of securing the AI supply chain
• The challenges of securing AI-assisted development

And stay tuned here to see the next blog and video in our Meet the C7 Team series featuring Senior Manager Nye Prior!

‍