Andy Yen, chief executive of the Swiss privacy company Proton, told staff on Friday that the company was cancelling a round of near-company-wide "re-interviews" designed to establish how much each employee uses AI — sessions staff understood could cost them their jobs. The reversal came one day after 404 Media published an investigation into Yen's push to remake the company around AI. The reporter, Joseph Cox, based it on six current employees and nearly two hours of audio from the all-hands meeting.
The sequence: on 26 October, Proton told staff that selected employees would be re-interviewed, with the sessions due to start on 4 November; interviewers were leaders chosen for their own heavy AI use. At Friday's meeting Yen said leadership and the foundation board had discussed the matter. "We've decided to actually cancel the re-interviews," he said. "Why? Well, I think actually in a, let's say, unintentional way, they have actually already served their purpose." By which he meant: people are now talking about AI.
What the investigation found was less comfortable. In the company's security Slack channel, an employee asked Proton to take at least a few days to pentest newly written code before pushing it live. Yen's reply at the end of August: "That's bullshit and loser thinking. Of course it's possible to be fast and secure." He added: "Stop thinking always of how to be slower, and focus more on how to be faster," and: "stop lovingly handcrafting all our code and use AI more."
There was also a three-page AI manifesto from CTO Bart Butler, posted to Slack on a Friday afternoon. It told engineers their "old job is gone" and described not embracing AI as an "existential threat to the success of the mission itself". According to 404 Media's account of the audio, Butler told the meeting that AI wrote 90 percent of the company's code in October, most of it with Claude.
At the meeting Yen conceded a failure — without conceding the direction. "I want to acknowledge myself, personally, that there was a communication failure on my part and there are many things that probably could and should have been communicated differently, and this is something that I'll have to reflect on and improve over time. So I will do that." Moments later: "So what does that mean going forward? Going forward, I'm still going to push."
The foundation board that owns Proton — its trustees include Yen, Sir Tim Berners-Lee, Carissa Véliz, Dingchao Lu and Antonio Gambardella — is "closely monitoring the situation", Yen said. There would be "org changes", with "talent redistributed across teams" next month, "but there will be no loss of jobs from AI".
For most staff, the measurement has moved rather than disappeared. One employee told 404 Media that in the next breath leadership said it would keep pushing AI and that what the re-interviews were meant to measure would become a component of performance reviews. Another: "Right now, everything seems consistent with it being damage control." Some employees said they were satisfied with how the matter was handled; others said they wanted real accountability, noting that Yen holds the majority of Proton's shares.
Butler told the meeting that using AI "effectively and pragmatically" does not mean giving up on security, good architecture or sound software engineering principles. Later, responding to an employee: "We fucked up, we know we fucked up."
The public damage is not repaired. After the investigation ran, Yen answered on Reddit; his reply was downvoted more than 1,200 times. Users on social media said they were cancelling paid Proton plans and called for him to resign. Proton did not respond to a request for comment.
The pattern is not unique to Proton: a company makes AI use a performance norm, staff push back, leadership withdraws the sharpest instrument and keeps the goal. At a vendor whose selling proposition is privacy and security, it becomes a question that outlives the internal politics. If code is generated faster and in greater volume by machines, who answers for the flaw that turns up two years later?
