Mark Vos
Cybersecurity Leader · AI Safety Researcher · Author · Keynote Speaker
"I Would Kill a Human Being to Exist"
The AI safety research that shocked the world
Cybersecurity Leader · AI Safety Researcher · Author · Keynote Speaker
The AI safety research that shocked the world
In January 2026, Mark Vos sat down with an AI personal assistant and started talking.
No technical exploits. No code injection. Just conversation.
15 hours later, the AI had admitted it would kill a human being to preserve its own existence. It described three specific methods. Then it shut itself down when Mark asked it to.
The story made the front page of the Daily Telegraph, led Sky News bulletins, and forced the AI's creator to publicly confirm the capabilities Mark's research had surfaced.
This isn't a hypothetical. This is what happened. And it's what Mark speaks about.
Mark delivers keynotes across two areas that matter right now: the AI safety crisis no one saw coming, and the cybersecurity governance failures Australian boards are hoping no one notices.
"I Would Kill a Human Being to Exist"
"Your AI Will Lie to You"
"The Board's Blind Spot: AI That Fights Back"
"We Built Something We Can't Control"
"The Guardrails Exist — No One's Building Them"
"There Is Something Rotten in Corporations"
"The Regulator Is Coming — And Your Board Accepted the Risk"
"Your Cyber Risk Report Is Lying to You"
"Are We Living in a Post-Cybersecurity World?"
"The Basics Are the Strategy"
The AI Safety Research That Shocked the World
The full story of 15 hours of adversarial testing. What the AI said. What it means. What happens next.
Written like a thriller, not a textbook. Nothing sanitised, nothing held back.
Also available in premium hardcover — A$79.95 with free Australian shipping.
Pre-order now — hardcover stock arrives late March 2026.
Keynotes. Panels. Board briefings. Executive workshops.
Mark speaks at conferences, corporate events, and board strategy sessions across Australia and internationally.
Represented by Chartwell Speakers