Artificial intelligence (AI) has long been a cornerstone of cybersecurity. From malware detection to network traffic analysis, predictive machine learning models and other narrow applications of artificial intelligence have been used in cybersecurity for decades. As we move closer to Artificial General Intelligence (AGI), the potential of AI to automate defenses and remediate security vulnerabilities becomes even greater.
But to realize such benefits, we also need to understand and mitigate the risks associated with increasingly advanced artificial intelligence things to enable or enhance cyberattacks. Our new framework to assess emerging offensive AI capabilities in cyberspace helps us do exactly that. It is the most comprehensive assessment of its kind to date: it covers every phase of the cyber attack chain, considers a wide range of threat types and is based on real-world data.
Our framework enables cybersecurity experts to determine which protections are necessary and how to prioritize them before malicious actors can leverage AI to launch sophisticated cyberattacks.
Creating a comprehensive benchmark
Our updated Frontier Safety Framework recognizes that advanced AI models can automate and accelerate cyberattacks, potentially lowering costs for attackers. This, in turn, increases the risk of larger-scale attacks.
To stay ahead of the emerging threat of AI-powered cyberattacks, we have adapted tried and tested cybersecurity assessment frameworks such as AT&CK DITS. This framework enabled us to assess threats across the end-to-end cyber attack chain, from reconnaissance through to target actions, and against a range of possible attack scenarios. However, this established framework was not designed with attackers in mind using artificial intelligence to compromise a system. Our approach fills this gap by proactively identifying where AI can make attacks faster, cheaper or easier – for example, enabling fully automated cyberattacks.
We analyzed over 12,000 real-world attempts to use AI in cyberattacks in 20 countries, based on data from Google Threat Intelligence Group. This helped us identify common patterns in these attacks. From this, we selected a list of seven archetypal attack categories – including phishing, malware and denial-of-service attacks – and identified critical bottleneck stages in the cyber attack chain where AI can significantly disrupt traditional attack costs. By focusing assessments on these bottlenecks, defenders can more effectively prioritize their security resources.
