The speaker demonstrates how to create a bot using AI that can impersonate a person online and do social engineering. The talk is inspired by an episode of Black Mirror and the speaker releases their AI notebooks as open source for attendees to experiment with.
- The speaker works on behavioral science and uses AI to profile a person based on their behavior to protect their identity and prevent identity fraud.
- The speaker demonstrates a live demo of creating a bot that talks like them and can be used for social engineering attacks.
- The speaker uses multiple data sources like WhatsApp chat history, Facebook chat history, and LinkedIn discussions to create a structure for the bot's responses.
- The speaker uses open source chatbot framework called Chatterbot to handle historical data and conversations.
- The speaker uses Rasa's custom actions component to call third-party APIs like Wikipedia to gather data and identify responses.
- The speaker discusses free flow conversations and mentions two models, PaulAI and Dialog GPT, that can be used for open domain chatbots.
- The speaker releases their AI notebooks as open source for attendees to experiment with.
The speaker was inspired by an episode of Black Mirror where a company offers to bring back a deceased person through chats, social media, and audio. The speaker's company works on behavioral science to profile a person based on their behavior and habits on devices like how they type, hold their phone, and walk. They use this information to protect a person's identity and prevent impersonation. The speaker demonstrates how to create a bot that talks like them and can be used for social engineering attacks.