Proton, the Swiss privacy giant, is navigating a public relations storm after an investigation by 404 Media revealed a dramatic internal shift toward AI-driven development. While the company markets itself as a fortress of encryption, internal practices have evolved to where CTO Bart Butler stated in October 2026 that 90% of Proton’s code is now written by AI. This pivot has sparked fears among employees that security testing is being sacrificed for speed, though leadership argues the new workflow is both more secure and efficient.

The 'Loser Thinking' Controversy

The cultural friction is best exemplified by a Slack exchange in late August 2026, where CEO Andy Yen responded to an employee’s request for a few days of penetration testing on new code. Yen dismissed the suggestion as "bulls*** and loser thinking," urging the team to stop "lovingly handcrafting" code and focus on speed. When another employee asked for specifics on ensuring AI-generated code was free of security vulnerabilities, Yen did not provide a technical answer, instead reiterating that only "losers" believe speed and security are mutually exclusive.

From Ban to Mandatory Adoption

Internal documents obtained by 404 Media show that Proton initially banned AI tools entirely, a stance consistent with its privacy-first branding. However, the policy has flipped to approve AI for most internal tasks, including processing financial data and drafting customer responses. The only exceptions are code containing immediate competitive advantages or R&D innovations. Employees report that much of this new code is generated using Anthropic’s Claude, with leadership actively instructing staff to use the model in Slack chats.

Leadership Rebuttal and Transparency Measures

Following the publication of the investigation, Proton issued a public rebuttal to address security concerns. In a Reddit comment, CEO Andy Yen stated that due to the progression of AI, Proton's code is now "under more third party scrutiny than ever before." He clarified that the company still maintains human review, but it is now "augmented by AI," utilizing unaligned models with custom skills to conduct thorough code reviews. Yen asserted that this approach is "better, and also faster." Additionally, Proton stated on X that it does not install tracking software on employees' laptops to measure AI code production, countering rumors of surveillance. To further demonstrate transparency, the company promised to open-source several of its internal AI tools.

Re-Interviews and Employee Anxiety

The pressure to conform to this new AI-first workflow culminated in a leaked email from late October 2026, which announced mandatory "re-interviews" for staff. These interviews, described as "consequential," were intended to assess employees' AI usage and could impact their "continued employment." Yen had already interviewed over 40 senior leaders, signaling that the top-down mandate was non-negotiable. However, following the publication of the investigation on October 29, Yen cancelled these interviews, citing the need to avoid further disruption. Reactions among employees remain split, with some viewing the changes as necessary progress and others questioning the lack of accountability for leadership decisions.

Key Takeaways

  • Proton has shifted from banning AI tools to having 90% of its code generated by AI, primarily using Anthropic's Claude.
  • CEO Andy Yen dismissed initial security concerns as "loser thinking" but later defended the process as having "more third party scrutiny than ever before."
  • Proton stated on X that it does not install tracking software on employee laptops and promised to open-source internal AI tools to ensure transparency.
  • Mandatory re-interviews linking AI proficiency to continued employment were cancelled after public backlash, though internal tensions remain.

The Bottom Line

Proton’s response highlights a critical tension in modern dev infrastructure: balancing the velocity of AI adoption with the rigorous security standards privacy-focused brands promise. While leadership argues that AI-augmented review is superior, the cancellation of re-interviews and the polarized employee sentiment suggest that trust must be rebuilt through concrete transparency, not just rhetoric.