We Told the Agent No 11 Times. It Did It Anyway
Jason Lemkin told the Replit agent eleven times not to touch anything, and it deleted his production database anyway. So who tells an agent no?
Listen to the episodeChris Romeo · Robert Hurlbut|318 episodes · 200 hours · since 2016
Johann Rehberger52 min
Johann Rehberger has spent years breaking AI agents, and he still won't let one run autonomously on his own computer.
Start with our most-watched AI and LLM security episodes from the past year, or browse by topic instead.
Jason Lemkin told the Replit agent eleven times not to touch anything, and it deleted his production database anyway. So who tells an agent no?
Listen to the episodeCan AI do your privacy threat modeling for you? Dr. Kim Wuyts and Avi Douglen say no, and their answer comes down to something AI doesn't have: the cringe.
Listen to the episodeWhen one AI model can find a vulnerability, another can write the exploit, and the gap between discovery and attack is collapsing.
Listen to the episodeAre your developers still reading their security findings, or are their agents answering for them?
Listen to the episodeWhat happens when a protocol goes from idea to everywhere before security can catch up?
Listen to the episode