ScamBack | Devpost

Call a scammer
Call in-progress
Call summary

Inspiration

We were inspired by YouTubers like Kitboga and Jim Browning, who dedicate their platforms to fighting scammers and helping individuals who fall victim to scams. Their efforts not only raise awareness about the prevalence of scams but also provide a unique form of justice by wasting scammers' time and resources. We wanted to take that idea further by automating the process and scaling it through AI.

What it does

ScamBack starts by contacting known scam phone numbers from a public database. Once connected, the app impersonates one of the scam victims the scammers typically prey on. Through AI-generated dialogue and realistic voice synthesis, it keeps the scammer engaged in conversation, wasting their time and preventing them from targeting real victims. The app continuously adapts to the scammer's responses, creating a dynamic and unpredictable experience that makes it harder for scammers to detect that it's not a real person.

How we built it

The application is built using Twilio to power the voice calls. There is a flask backend that acts as a WebSocket to receive information from the phone call and then send it to an LLM and TTS to generate the responses back to the scammer. The frontend is built with React.

Challenges we ran into

The most challenging aspect of this project was reducing the latency of the response to the scammer we are calling. We required our LLM and TTS to generate responses in 1-2 seconds in order for the response to sound human-like.

Accomplishments that we're proud of

We're proud that the bot actually works and can be used to waste a scammer's time so that they aren't scamming actual people.

What we learned

We learnt that most APIs for VoIP suck (I'm looking at you Vonage) except for Twilio. We also learned that there are very few options for fast TTS that actually sounds human-like so we are happy that we found one that we were satisfied with.

What's next for ScamBack

We would love to deploy our project so that it is always calling scam numbers and wasting their time so that there are less scammers scamming real people.