AI packing "Granny" to combat scammers and spammers

It's always nice to read about applications of AI for the greater good, so I loved this recent story of a fightback against scam and spam phone calls.  No-one likes a scammer (maybe apart from their Grannies!), so it was interesting to read of this AI approach to countering scammers and spammers from telecoms firm O2.  

The proposal is to set up an AI powered chat-bot to reply to spam and scam phone calls by deliberately mishearing or misunderstanding questions and wandering off-topic into chats about tea and cakes and long-winded reminiscing.  The idea is that the scammers get tied up in dealing with the bot until they lose interest, thereby reducing the time that they can spend trying to scam others.    

The approach solicited the scammers by publishing the phone number of web pages and other public sites.  However, it would be good if the approach could be extended to automatically detect unsolicited and unwanted calls and spring into action when needed.  

Although this is a somewhat unusual approach, computer implemented innovations in the field of cyber security are  often considered to provide a technical contribution at the UK-IPO and EPO, thereby distancing them from the “pure maths” or purely administrative exclusions from patentability.  As such, if you work in this area, then patentability might be something worth considering.