Featured inBloombergThe Wall Street JournalThe Telegraph
Fortinet asked Matthew to give the opening keynote at its System Integrator (SI) summit in Prague, where over three hundred of its top SIs and Managed Service Partners (MSPs) from around the world met to learn about the latest cyber threats and risks, and how to counter them.
The keynote was split into two parts, emergent cyber threats and emergent defence concepts. Its purpose was to open the audience's eyes to their unknown unknowns, so that they could be better prepared for the threats ahead as technologies improve exponentially and state-sponsored actors and hackers grow bolder.
"The purpose of the session was to open the audience's eyes to their unknown unknowns so that they, and all of us, can be better prepared for the threats that lie ahead."Matthew Griffin, Prague, February 2025
Matthew walked through the new tools attackers are developing to get around state of the art defences: using sound to watch your screen and clone your biometrics, using DNA malware to corrupt police forensics systems, and using human psychology to compromise, corrupt and destroy the AI foundation models and systems that hundreds of millions of people use every day.
He then turned to the development of autonomous agentic cyber offence and defence systems. He showed how white noise can corrupt and interfere with auditory attacks, how AI Large Audio Models (LAMs) and machine vision systems can see the latest polymorphic malware variants and phishing attacks, and how the latest predictive cyber defence systems are taking shape, among many other new concepts.
311 Intelligence
What this means for your leadership team
Assume attackers will use AI agents.
Matthew described autonomous agentic cyber offence alongside defence. Security teams should plan for attacks that adapt faster than people can respond.
Protect biometrics and AI models.
Sound can be used to clone biometrics, and psychology can corrupt AI foundation models. Both need their own defences and testing.
Invest in prediction, not just response.
Predictive cyber defence and AI that can spot polymorphic malware point to a shift from reacting to anticipating.
Three questions for your next leadership meeting
How would we respond to an attack run by an autonomous AI agent?
Which of our systems rely on biometrics that could be cloned?
Who tests our AI models against manipulation?
The trends and technologies behind this session
From the 311 Starburst, the 311 Strategic Intelligence Unit’s map of the technologies and trends shaping the next 50 years.
What do emergent cyber threats mean for organisations?
Attacks are becoming autonomous and are aimed at biometrics and AI models as well as networks. Defences need to become predictive and partly autonomous too.
How do Autonomous Cyber Reasoning Systems relate to this keynote?
They are AI tools that find flaws in software and prove they are real, part of the autonomous offence and defence the keynote described. The 311 Starburst expects wide use around 2032.
What is Adversarial Cyber Defence?
It spots attackers early and fights back as the attack happens. The 311 Starburst expects wide use around 2029.
How can 311 brief our leadership team on this?
Through a private executive briefing or leadership workshop, tailored to your sector, in person first.
Bring this to your leadership teamA private executive briefing or leadership workshop, tailored to your sector. In person first, anywhere in the world.
Analysis by the 311 Strategic Intelligence Unit, from Matthew Griffin’s session and 311 data.
If Matthew gave this session today
When AI slips beyond control, leaders must respond.
Rogue AI and The Unkillable Machine asks what happens when AI systems slip beyond human control, and what leaders need to do about it, a growing risk as autonomous agentic systems enter cyber offence and defence.
