A few days to test code before it goes live: that’s what a Proton employee asked for in August, and his boss, Andy Yen, responded with ‘loser mentality’. In internal messages revealed this week, the Swiss company behind Proton Mail is pushing its teams towards AI at a forced march, with interviews that could impact some employees’ jobs.
- Andy Yen qualifie de « loser thinking » la demande d’un employé qui voulait quelques jours pour tester la sécurité du code.
- Le CTO affirme que 90 % du code produit en octobre a été écrit par l’IA, sans audit.
- Des employés seront réinterrogés sur leur usage de l’IA, ce qui peut affecter leur emploi.
- Un document interne reconnaît que Proton n’atteindra pas son objectif de revenus récurrents annuels.
On Thursday, October 29, the American outlet 404 Media published an investigation by journalist Joseph Cox. It is based on eight current employees and one former one, as well as Slack conversations, meeting recordings, and a manifesto written by the CTO. When contacted, a Proton spokesperson did not dispute any element of the report.
Proton: A Security Testing Request Deemed ‘Loser Mentality’
According to messages reviewed by 404 Media, it all started with an exchange at the end of August in the dedicated security Slack channel. An employee asks to take at least a few days to do a pentest, meaning to attack newly written code oneself to find flaws before publishing it. Andy Yen replies:
“That’s bullshit and loser thinking. Of course it’s possible to be fast and secure.”
In other words: bullshit and a loser mentality, and of course it’s possible to be fast and secure. He adds that they need to stop thinking about how to go slower, then reiterates that “only losers think like that”.
Another employee points out to him that AI won’t magically solve problems because security is currently low priority. Andy Yen doesn’t address the substance. In front of the whole channel, he attacks another employee, summarizing their position as follows: it can be done, but by being even slower than today. Every developer, he writes, must better formulate their requests to the AI and check their code, like before by hand but faster. His directive: stop hand-holding all the code and use AI more.
Proton”: “90% of Code Written by AI”, Claims Its CTO
The CTO, Bart Butler, stated in a meeting that 90% of the code produced at Proton in October had been written by AI, according to a recording obtained by 404 Media. Several employees report that a large part of this code comes from Claude, Anthropic’s model. 404 Media adds that Andy Yen himself recommends using it in Slack conversations. The figure remains an unaudited oral statement.
In September, Bart Butler also posted a manifesto titled “AI @ Proton” on Slack. He writes that the engineers’ old job has disappeared, that not switching to AI represents an “existential threat to the success of the mission itself”, and that Proton must reach a point where most code is reviewed by AI rather than humans. Employees interviewed by 404 Media believe the text exaggerates what models can do today, with one calling it “ethically bankrupt”.
Other measures reported by 404 Media: Andy Yen created an email address for employees to report anything slowing AI adoption, so that management could, in his words, “ruthlessly crush” anything standing in the way of AI speed. Proton also installed software on computers tracking the amount of code generated by AI. Andy Yen finally wrote in Slack that 2026 would be the last year the company hires engineers: next year, it will recruit what used to be called product managers, tasked with building with AI.
Proton: Re-interviews on AI Usage That Could Impact Employment
On Monday, October 26, an internal email announced that some employees would be “re-interviewed” about their AI usage between November 4 and 11. The email, which 404 Media first revealed and which Proton later published in full on Reddit, according to Tom’s Guide, warns that these interviews will have consequences: they can impact roles, responsibilities, and job retention. Not everyone is summoned, and the interviewer comes from another team, among managers deemed strong on AI. Andy Yen has already interviewed over 40 engineering and product managers, according to the email.
Proton defends itself publicly on Reddit: the company says it is not conducting layoffs using AI as a pretext, claims it lacks talent and is recruiting across the organization. Internally, according to exchanges reviewed by 404 Media, the tone is harsher. A senior employee suggests identifying those falling behind to train them. Andy Yen replies that management already knows “more or less who is behind” and that, if they had given up on them, they wouldn’t bother with this process. When the employee tells him this rhetoric is based on fear, the boss retorts in front of everyone that if he’s accusing him of making a finance-driven decision without trusting him, “maybe this isn’t the right place for you”.
Proton Not on Track to Reach Its Annual Revenue Goal
An internal document obtained by 404 Media describes the situation. “Due to insufficient velocity”, Proton is not on track to reach its annual recurring revenue goal by year-end, meaning revenue secured by subscriptions. Growth is “far from what the market currently considers good”, and “by all metrics that matter”, the document writes, Proton is “an underperforming company”. Andy Yen, however, assures employees that the decision is not financially driven.
Proton in Public: A Privacy-Respecting AI, but Worried Employees
In public, the tone is more nuanced. In an interview with WIRED in August, Andy Yen said he was building AI tools because, if privacy companies don’t, the future belongs to the tech giants. The same company launched its Lumo chatbot in July 2025, then a version 2.0 in June 2026.
The Proton spokesperson acknowledges that “some people at Proton are unhappy with our direction, or don’t understand it, and many are worried”. In their determination to move fast, he says, leaders have sometimes been too abrupt or too slow to explain themselves, and they promise to work on it. The re-interviews begin on November 4.
