Digi International states frontier and agentic AI models are accelerating vulnerability discovery and multi-stage attacks.
“Frontier AI models offered by major developers, as well as various open-source models that are more difficult for governments to regulate, are increasingly capable of automating vulnerability discovery, accelerating the development of exploits, and identifying security weaknesses across complex product portfolios at a speed and scale that was not previously achievable. Nation-state actors and other sophisticated threat actors may also use these AI tools to enhance or automate additional attack vectors, including AI-generated phishing and social engineering campaigns, deepfake-based impersonation, and agentic AI systems capable of executing multi-stage attacks, such as ransomware or extortion campaigns, with limited human involvement. We believe this represents a material change in the threat environment facing many companies, including those like ours that develop and sell networked hardware and software products.”