The Globe and Mail reports in its Thursday edition that University of Toronto researchers have identified a new cyberthreat involving artificial intelligence that can create a self-modifying computer worm. Alexandra Posakzki writes that cybersecurity concerns have focused on large language models like OpenAI's GPT-5 and Anthropic's Claude Mythos Preview, which were initially released to select groups due to the risks they pose if misused. On Tuesday, AI Minister Evan Solomon announced a partnership between the Canadian government and Anthropic to use Mythos for enhancing cybersecurity. Research from the CleverHans Lab at the University of Toronto and the Vector Institute suggests a cheaper and more accessible way for hackers to use AI for malicious purposes. Researchers say any device connected to the Internet is at risk, from laptops, to cameras, to printers. Worms spread automatically between machines without human intervention, copying themselves onto each device. Traditionally, they followed human-generated scripts and would fail against defenses they weren't designed to bypass. AI can enhance worms by enabling them to adapt in real-time, creating tailored attack strategies for each machine they encounter.