There’s been a lot of science fiction around the dangers around AI in American culture. However, I think the real dangers are actually something different. What about malicious use from those who want to use the interface for illegal activity? Is this being addressed and how so if anyone knows? Things like drugs and sex might be in future discussions for some users, especially among the young who might use the interface in ways that can’t be truly predicted. Any thoughts?
I agree with your perspective. In my view, the primary risk isn’t that AI becomes independently dangerous—AI isn’t conscious or autonomous in the way science fiction often portrays it. It’s a tool, and like any powerful tool, it can be used for both beneficial and harmful purposes.
The bigger concern is how people choose to use it. Misuse for illegal activities, fraud, misinformation, or other harmful purposes is a real challenge, and that’s where much of the focus should be.
This is already being addressed to some extent. For example, the European Union has introduced AI regulations aimed at promoting the safe and ethical use of AI. There are also cybersecurity frameworks and legal efforts designed to reduce risks such as misuse, bias, discrimination, lack of transparency, and misrepresentation. While no system can eliminate every risk, there is a growing international effort to ensure AI is developed and deployed responsibly.
Do you think there might be a problem with small party AI that can jailbreak safety guardrails that most large use AI has in place? Should that be legal?
That’s a reasonable concern. One important thing to keep in mind is that the AI ecosystem isn’t a single monopoly or one unified system. While there are a relatively small number of major AI companies, they are independent competitors with different models, architectures, and safety mechanisms.
Because of that, there isn’t a single “jailbreak” that would compromise the entire AI landscape. However, individual models or frameworks can sometimes be attacked or bypassed, just as software systems, networks, and other IT infrastructure can be targeted by cybersecurity attacks.
This isn’t unique to AI—it’s an ongoing challenge across information technology. The goal is to continually improve security, safety guardrails, and monitoring as new vulnerabilities are discovered.