Prompt hacking = phone phreaking?

I can’t help but notice how similar prompt injection attacks are to phone phreaking. Both rely on the control signal being on the same medium/conduit as the content. For phones, it was making the right frequency and duration tones to take control, along with social engineering to get access. Prompt injection relies on similar techniques, just with an LLM as the target for the social engineering and using tricks to escape user-land for prompts.

If you haven’t read Ghost in the Wires, it’s an interesting read about a bygone era with social engineering implications that reverberate today: https://a.co/d/7oi5AVd.