Vancouver Cybersecurity Conference

Every organization has a layer in its security stack that nothing is watching: the running application. Until recently, the cost and complexity of exploiting it at scale provided accidental protection. Mythos ended that. This session uses Anthropic’s AI vulnerability research system as a lens to examine the application layer threat landscape, why the industry’s go-to controls were never designed to see these attacks, and what defense-in-depth must look like now.

Mythos-class frontier AI models have generated a lot of fear, uncertainty, and doubt. The hype, the headlines, and the hand wringing have been plentiful, but where does the truth lie, and what should organizations do to protect themselves in this accelerated bug finding landscape? This talk will give the inside scoop from a security vendor’s own internal security team. What are these models good at, what will it mean for commercial software versus open source? Will patching faster be possible, or even good enough? We’ll share how we are using these tools to secure our products, as well as how we are securing our own IT environment in this brave new world.

AI is quickly becoming the most disruptive technology since cloud. But It can be used either  for good or bad. Cybersecurity teams are facing an increasingly intense threat landscape, as attackers enhance the sophistication, scale, and speed of exploitation making use of AI assistants and AI agents. This presentation covers some of the main challenges for cybersecurity teams: Frontier AI models (Mythos) able to reveal and exploit a huge number of zero-day vulnerabilities, defying the established vulnerability management processes;  AI is transforming the attacks so there is the need to use automation and agentic AI to detect and respond to incidents at machine speed; and how to protect AI models, AI applications and chatbots, and defend against the loss of data or prompt abuse.

This session profiles the landmark DDoS event of October 31, 2024, which signaled a paradigm shift in threat landscapes with the 2025 emergence of the Aisuru botnets and the 2026 Mythos exploit framework. We provide a technical breakdown of how this TurboMirai-class malware weaponized Android TVs and IoT hardware to achieve unprecedented hyper-volumetric scale. By analyzing this case study, we examine the rising trend of terabit-scale assaults favored by both state-sponsored actors and decentralized threat collectives.