Proton's CEO Cancels AI 'Re-Interviews'
Proton’s now-cancelled plan to “re-interview” staff about their AI usage, combined with the CEO’s reported comments dismissing security concerns as “loser thinking,” is alarming many users who chose the service for its privacy-first positioning. Commenters question whether a company that pushes speed and AI-generated code over pentesting can still credibly sell itself as a secure alternative to Google, with some cancelling subscriptions and weighing options like Fastmail, Tuta, self-hosting, or smaller niche providers. The episode is framed as part of a broader pattern of AI hype, executive overreach, and erosion of trust in tech leadership.
Overall reaction to Proton’s AI push and “re-interviews”
- Many see the now-canceled AI “re-interviews” and internal comments (e.g., dismissing security concerns as “loser thinking” and urging AI-generated code over “handcrafted” code) as a major cultural red flag.
- Several commenters describe this as reputationally suicidal for a company whose brand is “privacy and security first.”
- Some appreciate the bluntness: if the company wants to be AI‑centric, better to say so openly and let misaligned staff leave. Others argue that honest delivery doesn’t excuse contemptuous tone or security de‑prioritization.
- The CEO’s follow-up statement is widely viewed as a non-apology focused on “communication failure” rather than acknowledging that the underlying stance was wrong.
Security, trust, and Proton’s positioning
- Core concern: a CEO pushing speed and AI over security reviews contradicts the company’s privacy/security marketing.
- Commenters question whether Proton is still a “security company” or just chasing generic AI hype.
- Some reference negative engineering-culture reviews predating AI, suggesting this is part of a longer pattern.
- A few remain cautious, wanting corroboration from more outlets or concrete product/security regressions before switching.
Customer reactions and alternative providers
- Many paying customers say they have already canceled or plan not to renew. Others are locked in short-term by recent renewals or migration friction.
- Common advice: buy your own domain to avoid lock-in and make provider changes easier.
- Alternatives discussed:
- Fastmail: widely praised for reliability and UX; noted to use AI for support and provide an MCP API, but not to pipe user mail through AI by default. Concerns: Australian company, US/EU data regions, subject to US/Five Eyes influence.
- Gmail: strong security track record but ad-based and US big-tech.
- Zoho, Tutanota, mailbox.org, small indie hosts, and future Mozilla “ThunderMail” are mentioned with various tradeoffs.
- Several note that email is inherently hard to secure end-to-end; providers differ mainly in practices, incentives, and jurisdiction.
Debate over AI in engineering
- Some argue that refusing AI makes formerly strong engineers uncompetitive, likening it to refusing CAD in architecture.
- Others strongly reject this, saying:
- Prompting LLMs is not a deep skill.
- Overreliance causes skill atrophy and opaque systems.
- “Productivity” metrics (lines of code, commit counts) are misleading.
- For a security-focused company, commenters stress that AI should enhance quality and assurance, not justify skipping pentests.
Broader critique of tech leadership and culture
- Many see this as part of a larger pattern: “AI psychosis,” extreme growth-at-all-costs capitalism, and normalization of abusive “tech bro” leadership.
- Several emphasize that demanding performance doesn’t require insults; treating staff as “losers” is seen as both unethical and strategically self‑defeating, driving away exactly the kind of security‑conscious talent Proton most needs.