Protecting the elderly from scam callers with GPT AI?

I am interested in brain storming ideas with the hacker news community about how Open AI technology could be use to protected the elderly from scam callers. Scam callers operate with near impunity in many parts of the developing world and often are targeting the most vulnerable and naturally trustful members of our societies. It was estimated 8.8 billon is stolen a year mostly from our elderly parents and grandparents who are in their most venerable stages of life. Something I have been thinkin about lately is how the success of these scammers largely rely on the self filtering out of victims, by that I mean most people don't answer unknown calls, or if they do they quickly hangs up when they hear the robotic voice that tries to get some basic info. (Often these scams deploy a "robotic dialer" whose main purpose seems to be to insure there is a human on the other side of the line willing to talk before transferring them to the relatively expensive call center workers.) It seems to me that if everyone who received these purposely answered them and tried to waste as much of the callers time as possible the entire model would be unprofitable and unsustainable. Obviously know one wants to spend their day answering and wasting scammers time (actually lots of people do see Jim Browning's Channel on youtube: https://www.youtube.com/channel/UCBNG0osIBAprVcZZ3ic84vw ) but not enough people would be able to spend the time need to do this.

This is where OpenAI powered technology could come in, I am envisioning an app that would allow people to send incoming scam callers to AI powered chat bot that is optimized to waste as much time as possible, and setup VoIP phone numbers setup to send to the system.

What do you guys think some of the (technical, legal, etc..) roadblocks are of such an idea. Obviously a lot of unwanted but not (strictly) fraudulent calls would be caught up in this, think auto insurance calls that got your phone number from some website etc, does this open the project up to legal issues? In my opinion there is actually a lot of gray area as to what constitutes a scam. Is a company that is cold calling and using high pressure sales and scare tactics to convince an elderly person they need a relatively useless "Identify Protection" service to protect them from the scary scammers, scamming them or not? I assume with out some sort of approval from them this would violate OpenAI api policies? What do you guys think of my "Operation Elder Shield" idea any feedback is appreciated?

Thanks for any feedback.