It wasn't AI that flipped Proton's charge

The thread is less interested in whether Proton should use AI at all than in whether an AI-driven push for speed is incompatible with a privacy business whose entire value rests on caution. Commenters repeatedly return to the reported incident in which security testing was treated as an obstacle: for an encrypted email provider, pentesting and deliberate release cycles are not bureaucratic drag but the product customers believe they are buying. Many see the CEO’s rhetoric and reported AI-related pressure as a self-inflicted trust crisis, made worse by prior worries that Proton has chased feature expansion over polish. A smaller group argues that the backlash is racing ahead of the evidence. They find the internal AI guidance more measured than the surrounding coverage suggests, use coding assistants themselves, and reject the idea that AI-authored code is automatically reckless. One forceful pro-AI view casts resistance to AI-assisted development as denial of tools that can aid programming and security work; even supporters of that view concede that expert review remains essential. The practical consensus is narrower: AI can improve developer productivity, but it cannot excuse cutting validation, weakening privacy safeguards, or treating careful security engineering as a liability. Customers who chose Proton specifically to escape Big Tech say this kind of AI rollout gives them a reason to prepare an exit.