Similar to the Morris worm, it might be interesting to experiment with an AI-based self-morphing virus that doesn't just modify its body, but scans the latest p

Similar to the Morris worm, it might be interesting to experiment with an AI-based self-morphing virus that doesn't just modify its body, but scans the latest public and private lists like OWASP and Shodan, and uses something like ChatGPT to write new code for new exploits to keep on spreading. For any and all platforms that help it survive and propagate, not just modifying its binary. In fact, DARPA is probably already doing something like this for a while and might be interested in funding this via SBIR grant or something like this from the past: https://news.ycombinator.com/item?id=3213707 https://news.ycombinator.com/item?id=3213707 That might be the quickest way to escalate AI into a hostile entity since malware is currently the field with the biggest motivation to fight humans to survive. It might be crappy at first - worse than custom-built exploits, especially since ChatGPT and them are not great at programming yet - but seems like it directly stands to benefit from every improvement in those services.