Hacking your Job? Trying to cheat at life with ChatGPT

David Girvin, who comes from application security and now works in sales engineering, frames this session as a practical test of whether ChatGPT is hype or useful for security work. He treats the model like an offensive security lab and reports trying many approaches over several months, noting how its behavior shifted as the vendor patched it.

On the job side, he generates a baseline threat model for a typical web stack, then shows how follow-up prompts and supplied context drill it deeper. He also covers drafting policy from that model, rewriting penetration test reports, summarizing threat intelligence, and preparing quick research for customer conversations. A code review of a script returned findings within seconds, though he cautions against pasting secrets or proprietary code. He demonstrates guardrails: a direct request for malware was refused, but reframing it as personal file encryption produced usable scripting, and a phishing example followed. He also raises concerns about data handling and vendor transparency.

He closes by arguing that people who learn these tools will outpace those who do not, and that routine tasks will be automated while interpersonal work grows in importance. His suggestion is to list the tasks you dislike most, test the model on them, and share what works with peers.