---
title: "Proton Scrapped Its AI Loyalty Checks in Four Days, Kept the Metric"
description: "The CEO who called pentesting ‘loser thinking’ folded the same AI-use criteria into ordinary performance reviews instead"
author: "Luis Goa"
published: 2026-11-02T14:52:58Z
modified: 2026-11-03T03:21:45Z
url: https://rews.cc/a/proton-scrapped-its-ai-loyalty-checks-in-four-days-kept-the--6663b8
language: en
tags: ["proton", "ai", "security", "workforce", "open-source", "tech"]
publisher: "Rews (https://rews.cc)"
---

# Proton Scrapped Its AI Loyalty Checks in Four Days, Kept the Metric

*The CEO who called pentesting ‘loser thinking’ folded the same AI-use criteria into ordinary performance reviews instead*

By Luis Goa · November 2, 2026 · https://rews.cc/a/proton-scrapped-its-ai-loyalty-checks-in-four-days-kept-the--6663b8

## In brief

- Proton announced AI ‘re-interviews’ on Oct 26 that could cost staff their jobs, then cancelled them on Oct 30
- CEO Andy Yen had called a request to pentest new code before shipping “bullshit and loser thinking” in an August Slack exchange
- CTO Bart Butler said AI wrote about 90% of Proton’s code shipped in October, per 404 Media’s reporting
- The AI-use criteria from the cancelled re-interviews will instead be folded into routine performance reviews
- A Proton forum petition for Yen to resign drew 2,020 votes; his Reddit defense was downvoted over 1,200 times

Four days. That’s how long it took Proton to go from announcing a company-wide round of ‘AI re-interviews’ that could cost employees their jobs to cancelling it. On October 26, an internal email told staff the interviews, scheduled for November 4 to 11, were “consequential” and could lead to “changes in roles, responsibilities, and continued employment,” as [Joseph Cox reported for 404 Media](https://www.404media.co/proton-to-re-interview-staff-about-ai-to-see-who-gets-to-keep-their-job/). By October 30, CEO Andy Yen had killed the plan, as [404 Media’s Samantha Cole reported](https://404media.co/reaping-and-sowing-ceo-edition/).

The mechanism itself is the interesting part. According to the leaked email, Yen had already personally run the process on more than 40 senior leaders across engineering and product before extending it to almost everyone else. The stated purpose was to find out how staff were using AI in their jobs; the stated consequence, in the company’s own words, reached as far as continued employment. [rews covered the original announcement](https://rews.cc/a/proton-will-re-interview-employees-about-ai-to-decide-who-ke-cae143) when the email first leaked; Proton’s public line at the time, posted to Reddit, was that it was “not doing layoffs and using AI as an excuse” and was in fact hiring, citing a talent shortage.

The number behind the push is concrete, if unaudited: Proton’s CTO, Bart Butler, told staff AI wrote about 90% of the company’s code shipped in October, according to the follow-up investigation 404 Media published on October 29, based on interviews with eight current and one former employee plus Slack logs and meeting recordings. Butler had put the company’s AI strategy in writing in an internal document titled “AI @ Proton, Engineering Manifesto,” arguing Proton needs to reach a point where most code is reviewed by AI rather than humans. The manifesto also concedes the technology isn’t there yet, and that humans remain, in Butler’s own words, “much less prone to hallucination and being misled by bad or malicious inputs” than models are. That is the document Yen’s rhetoric was supposed to be executing.

![Two robotic arms write red blocks and inspect a crack through a magnifying glass, while a human hand reaches toward the work surface.](https://rews.cc/img/cfbcccc00e7b5e1bf31f66bae79f462cbf0b6062.webp)

The rhetoric is where the backlash actually started. In late August, a Proton security engineer asked, in the company’s security Slack channel, for a few days to pentest newly written code before pushing it live. Yen’s reply: “That’s bullshit and loser thinking. Of course it’s possible to be fast and secure.” He followed up: “Stop thinking always of how to be slower, and focus more on how to be faster,” and, further down the thread, “only losers think like that.” His broader instruction to engineering was to “stop lovingly handcrafting all our code and use AI more.” Separately, Yen set up a dedicated email address for staff to flag anything slowing AI adoption, so leadership could, in his words, “ruthlessly smash everything that stands in the way of AI velocity,” and installed software on employee machines to track AI-generated code volume. He told staff this would be the last year Proton hires engineers in the traditional sense; a later message to staff, per 404 Media, read simply: “Your old job is gone.”

An internal document reviewed by 404 Media supplies the business case for the urgency: Proton is off track for its annual recurring revenue goal, attributed internally to “insufficient velocity,” with year-over-year growth “far away from what the market today considers to be good.” The document’s own verdict: “By all metrics that matter, we’re an underperforming company.” Proposed fixes described in the same reporting included an “elite fast team” drawn from 5 to 10% of staff with a 30% bonus attached, and a consideration, later dropped, of relocating headquarters to the US or China before Swiss privacy law was cited as the reason to stay.

What that produced inside the company, per 404 Media’s sourcing, was less a debate than a chill. One employee described a split persona: “Whatever persona Andy has out of Proton, he’s a straight up bully inside. He routinely breaks his little facade in Slack and calls people losers and stupid, often for stuff he’s paying them for.” Another, more bluntly: “The CEO is in full AI psychosis and has been for a while now.” A third summed up the atmosphere: “A fear based culture feels like a cancer to me.”

**How the re-interview plan unravelled** Timeline of events, August to October 2026

| From | To | How |
| --- | --- | --- |
| Security engineer requests pentest time *(Slack, late August)* | Yen: 'loser thinking' *(rejects the delay)* | flags risk |
| Yen: 'loser thinking' *(rejects the delay)* | 404 Media investigation *(Oct 29, 9 employees cited)* | quoted in report |
| Re-interview email leaked *(Oct 26, jobs at stake)* | 404 Media investigation *(Oct 29, 9 employees cited)* | leaked email |
| 404 Media investigation *(Oct 29, 9 employees cited)* | All-hands meeting *(Oct 30, ~2 hours)* | forces response |
| All-hands meeting *(Oct 30, ~2 hours)* | Re-interviews cancelled | announced |
| Re-interviews cancelled | Criteria moved to performance reviews | metric kept |

![How the re-interview plan unravelled. Timeline of events, August to October 2026](https://rews.cc/img/baaace6937daf9a463f0114d96af20ad095613ce.webp)

Based on [404 Media](https://www.404media.co/manifestos-fear-and-loser-thinking-inside-protons-ai-reckoning/)

The diagram above is the whole sequence: a dismissed security request becomes a quoted embarrassment, which turns a leaked HR email into a published investigation, which forces a reversal that keeps the underlying metric and only removes the interview format around it. At the October 30 all-hands, audio of which 404 Media obtained, Yen told staff: “there was a communication failure on my part and many things could have been communicated differently,” and “we’ve decided to actually cancel the re-interviews.” He added a caveat that matters more than the apology: “Going forward, I’m still going to push.” Butler was shorter: “we fucked up, we know we fucked up.” The AI-use criteria the re-interviews were built to capture didn’t disappear; they move into ordinary performance reviews, a channel with the same stakes and a much lower chance of ending up in a reporter’s inbox.

Employee reaction split on whether that distinction matters. One source told 404 Media after the meeting, “I don’t think he will change anything,” framing the reversal as damage control in response to what another called a rebellion. Whatever the internal politics, structural accountability is limited by design: since June 2024, Proton’s controlling shareholder has been the nonprofit Proton Foundation, [as TechCrunch reported at the time](https://techcrunch.com/2024/06/17/privacy-app-maker-proton-transitions-to-non-profit-foundation-structure/), with Yen chairing its board of trustees alongside Tim Berners-Lee, Oxford ethicist Carissa Veliz and FONGIT director Antonio Gambardella. Yen has said no change of control can occur without the foundation’s consent — a structure built to block a hostile takeover from outside, which also means the board overseeing Yen is the board Yen chairs.

![A cream office chair holds a dark suit jacket beneath a coral articulated lamp shaped like an eye, against a bright blue background.](https://rews.cc/img/cafac2643612f135ad97c3c7055bee913920bbd8.webp)

The public reaction has outlasted the policy it was aimed at. A proposal on Proton’s own feedback forum calling for Yen to step down had drawn 2,020 votes by the time Decrypt checked it, as [Decrypt reported](https://decrypt.co/382353/proton-users-revolt-ceo-calls-security-caution-loser-thinking-ai). Yen’s own defense of the AI push, posted on Reddit, was downvoted more than 1,200 times, [BigGo Finance reported](https://finance.biggo.com/news/ef189055-f427-4b43-8f9c-67a1b627f254), with many commenters saying they were cancelling paid plans. None of that is an audited churn number — a cancellation post documents anger, not a verified account closure — but it’s a visible tally on a company that, by its own team page, serves [more than 100 million user accounts with over 700 employees](https://proton.me/about/team). A Proton spokesperson told 404 Media “some people at Proton are unhappy with our direction, or misunderstand it, and many people are anxious,” and that the company has “sometimes been too blunt or too slow in explaining ourselves.”

Folding the re-interview’s criteria into a standing performance review doesn’t change what’s being measured, only who can see the measurement. The open question the backlash hasn’t settled is simpler than the governance chart: Proton sells encrypted mail, VPN and password storage on the premise that its engineering is trustworthy by construction. Its CTO says nine-tenths of its code is now written by a system his own manifesto says isn’t yet reliable enough to review itself. Its CEO called a request to pentest that code before shipping it “loser thinking.” Nobody involved has published a number showing which of those two claims the company’s security practice actually follows.

## See also

- [Proton AG](https://en.wikipedia.org/wiki/Proton_AG) — en.wikipedia.org · Background on Proton's corporate history, products and ownership
- [Analysis of Bart Butler's internal AI manifesto](https://flaviocopes.com/proton-ai-manifesto/) — flaviocopes.com · A developer's read of Proton's 'AI @ Proton' engineering document
- [Privacy Guides community thread on the investigation](https://discuss.privacyguides.net/t/manifestos-fear-and-loser-thinking-inside-proton-s-ai-reckoning/41837) — discuss.privacyguides.net · Privacy-focused users discussing the 404 Media findings
- [Proton's AI controversy is a trust crisis](https://goosed.ie/news/protons-ai-controversy-is-a-trust-crisis/) — goosed.ie · Argues the episode undermines Proton's core sales pitch
- [Proton Mail goes full AI slop](https://pivot-to-ai.com/2026/10/30/proton-mail-goes-full-ai-slop-tuta-starts-going-slop/) — pivot-to-ai.com · Compares Proton's AI pivot with rival Tuta's approach