Benchmarking the Security Capabilities of Large Language Models
Comparative Sophos X-Ops testing not only indicates which models fare best in cybersecurity, but where cybersecurity fares best in AI.
Publications Our peer-reviewed research
The Dark Side of AI: Large-Scale Scam Campaigns Made Possible by Generative AI
Generative artificial intelligence technologies such as OpenAI’s ChatGPT and DALL-E have created a great deal of disruption across much of our digital lives. Creating credible text, images and even audio, these AI tools can be used for both good and ill. That includes their application in the cybersecurity space.
Sophos AI team to present at CAMLIS
The conference on machine learning in cybersecurity is key to open exchange of research and knowledge.
SophosAI at DEF CON: Orchestrating large-scale scams using text, audio and image generative AI
AI Village talk highlights how generative can be used to automate the creation of fraud campaigns, generating hundreds of fraudulent sites.
And I Shall Call It Mini-Me GPT: Using Large Language Models to Classify the Uncharted Web
Sophos AI team employs GPT and other large language models as teachers to train smaller models to label websites.
GPT for you and me: Applying AI language processing to cyber defenses
Three SophosAI projects harness the model behind ChatGPT for better detection of malicious activity.
An open-source ML toolkit for automatically generating YARA rules
The SophosAI Artificial Intelligence team has developed a machine-learning based tool that generates YARA rules for detecting specific types of threats.
SophosAI presentations at Black Hat, BSides LV and DEF CON AI Village
Six presentations in Las Vegas this week cover multiple cybersecurity applications of machine learning.
Building the AI-Assisted SOC: Sophos’ Five-Year Perspective
Looking ahead to the Security Operation Center of the future, forged from developments in XDR, AI innovation, and programmable security posture and powered by the AI-UX value circuit.
Using Undocumented Hardware Performance Counters to Detect Spectre-Style Attacks
In this paper, we’ll first introduce our version of Spectre variant 4 with evasive changes that can bypass any detections using conventional cache miss, branch miss, and branch misprediction counters. We’ll then show how our model using select undocumented counters is able to detect this new edited variant, and how it is also able to detect a novel Spectre implementation submitted to Virus Total.