Proton CEO Andy Yen: Security testing is loser thinking
In the company’s security Slack channel, Proton CEO Andy Yen called an engineer’s suggestion that new code undergo at least several days of penetration testing before going live loser thinking, and said that they should stop always thinking about how to slow things down and think more about how to speed things up, then added that only losers would think this way, asking the team to stop carefully hand-coding and use more AI.
404 Media reporter Joseph Cox interviewed 8 current employees and 1 former employee, obtaining internal Slack records, meeting recordings, and an AI manifesto written by CTO Bart Butler. In a meeting, Butler said that by October 2026, 90% of the company’s code had been written by AI (some using Anthropic’s Claude); Yen said 2026 would be Proton’s last year hiring traditional engineers, and that starting next year the company would hire people who can build products with AI. The company has installed software on employees’ computers to track the proportion of AI-generated code, and plans to re-interview employees about their AI use; an internal email said the interview would be consequential.
Internal documents show that, due to insufficient speed, Proton did not meet its annual recurring revenue target as planned, with year-over-year growth far below what the market currently considers good, and said that by every important metric, it is a poorly performing company. Faced with detailed questions from the reporter, a Proton spokesperson did not deny the specific facts in the report, saying that some employees are dissatisfied with or misunderstand the direction, many feel anxious, and acknowledging that management has sometimes been too direct or too slow to explain things; it will improve communication in the future and help employees retrain.
Original source:
Title: Manifestos, Fear, and ‘Loser Thinking’: Inside Proton’s AI Reckoning Publisher: 404 Media / Joseph Cox Publication date: 2026-10-29 URL: https://www.404media.co/manifestos-fear-and-loser-thinking-inside-protons-ai-reckoning/ Nature of source: investigative report Verification status: Read the public summary and key original passages widely cited by multiple English-language media outlets; Proton did not deny the core facts Supporting content: Yen’s Slack remarks, Butler’s statement that 90% of code was written by AI, the hiring shift, re-interviews, internal revenue documents, and the company response
ABAB AI Insight
Proton built user trust through end-to-end encryption and an anti-big-tech positioning; the market differentiation of its core products, Proton Mail and Proton VPN, depends on a public narrative that security comes before speed. Yen publicly criticized AI as a surveillance tool, yet internally he called shortening the security testing cycle loser thinking, and this split between public and internal narratives directly affects the sustainability of its brand premium.
Butler’s direction that the old jobs no longer exist and that most code will in the future be reviewed by AI creates tension with the traditional requirements of privacy products for manual review and traceability. When the proportion of AI-generated code is tracked in real time by internal tools and tied to employee retention, the incentive structure for security review shifts from caution to speed, which may change how product vulnerabilities are found and fixed.
The company admits it missed its ARR target and attributes weak growth to insufficient speed. AI全面化 is positioned as an acceleration tool rather than a mere technical experiment. For a privacy service provider that relies on subscription revenue, the pressure on revenue and the reallocation of security processes determine whether capital flows into traditional engineering capability or into AI-driven product teams that can iterate faster.
Historically, the trade-off between security and speed has repeatedly appeared in encrypted communications companies: an excessive pursuit of iteration speed has led to audit vulnerabilities, while excessive caution has faced pressure from larger companies’ feature dominance. Proton’s public labeling of security testing as loser thinking turns this trade-off from an internal management issue into a user-observable trust variable. Its long-term impact will depend on actual product security incidents and the pace of user migration, rather than management statements themselves.
ABAB News · Cognitive Law
For a brand that prioritizes security, when security testing is called loser thinking internally, the source of its premium has shifted from product mechanisms to narrative management.
When the code generation ratio is tracked in real time and tied to positions, the incentive for review shifts from finding vulnerabilities to avoiding becoming a speed obstacle.
When missed revenue targets appear alongside forced AI acceleration, the technology choice first serves the path to capital returns rather than the continuation of the original security commitment.