Proton's CEO cancelled the AI re-interviews and kept the policy that produced them

On Friday, Proton CEO Andy Yen cancelled the company-wide AI re-interviews that had required employees to explain how they use the tool, with the plain implication that the answer could decide whether they kept their jobs. He announced the reversal at an all-hands meeting one day after 404 Media published the investigation that made the process public, and in the same meeting said he was still going to push. Six current employees and audio of the roughly two-hour call, both obtained by the outlet, describe a company that withdrew a procedure without withdrawing the policy behind it.

The procedure went; the target stayed

Yen's stated reason, according to the recording, was that the interviews had done their work by accident. "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." Leadership had spent the previous couple of days discussing the matter and had spoken to the Proton board, which he said is "closely monitoring the situation". He also described the reorganisation due next month, which involves "redistributing talent across teams" and, in his words, "no loss of jobs from AI".

What was cancelled is a process, and a process is the cheap thing to cancel. Days earlier, employees had been told in writing that "These re-interviews are consequential", which current staff read as a warning that the interviews could end a career at Proton. That email is what turned a management preference into a test with a pass mark. Nothing in Friday's meeting, as 404 Media describes it, revoked the preference.

Still pushing, and still measuring

Near the start of the meeting, Yen said:

"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."

"So what does that mean going forward? Going forward, I'm still going to push."

The two statements arrived minutes apart, and the second one answers the first. Employees heard it that way. One told 404 Media that after the cancellation, "They said in the next breath they want to continue pushing AI and that the things that were to be measured in the re-interviews would become a component of the performance reviews." The instrument changed; the measurement, on that account, did not. Another employee put it more bluntly: "They have said clearly the direction had not changed. As a Proton user that doesn't reassure me." The same person said the interviews "weren't cancelled until this became public despite concerns raised by many of us", and that "Right now, everything seems consistent with it being damage control."

Not everyone the outlet spoke to was hostile. One employee said they were "mostly happy with how this was addressed including by Andy". The split matters less than the sequence: a process that was going to help decide employment was defended in private, reported on Thursday and withdrawn on Friday.

The complaint was about shipping, not about interviews

At the end of August, an employee in the company's security Slack channel asked Proton to take a few days to pentest newly written code before pushing it live. Yen's reply, quoted in both stories, was: "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." His conclusion for the team was to "stop lovingly handcrafting all our code and use AI more."

That exchange is the substance of the story, and the re-interviews were downstream of it. CTO Bart Butler said in a meeting that 90 percent of code at Proton in October was written by AI, much of it with Anthropic's Claude, and his AI manifesto, posted to Slack, opened with "Your old job is gone" and called a failure to embrace AI an "existential threat to the success of the mission itself". It set out three changes, including getting Proton to the point where most code is reviewed by AI rather than by people.

The disagreement in that channel is not whether a model can write code. It is whether a few days of testing is worth the exposure it prevents, and Yen's reply treated the question as evidence of the wrong attitude. A pentest window delays a release; it does not reduce how much code a team can produce. Whether that delay is affordable is a judgement about risk, and the CEO made it in a public channel by calling the alternative loser thinking.

What an apology for communication leaves out

Later in the call, Butler said Yen "needs to, and he has committed to, doing better" on how he communicates with employees, and added that "Using AI effectively and pragmatically does not mean that we've given up on security, or good architecture, or sound software engineering principles." He also said, responding to an employee: "We fucked up, we know we fucked up."

All of that is an apology for delivery. Nothing in 404 Media's account of the meeting describes a change to the August guidance, a new requirement for human review, or a longer testing window before code ships. The one process that got cancelled was aimed at employees rather than at the code.

It is worth being precise about what has not been established. Neither article reports a breach, an exploited vulnerability, or a shipped defect traced to AI-written code at Proton. The security problem is a concern employees raised and the CEO dismissed in a Slack channel; nothing in either piece converts that concern into an incident. A spokesperson, responding to questions from 404 Media for Thursday's story, did not dispute the reporting, and said that in its determination to move fast the company had "sometimes been too blunt or too slow in explaining ourselves".

The customers were never the audience for this fix

Proton sells privacy, which makes its internal engineering culture a consumer product whether or not management treats it as one. In the 24 hours after the investigation, many Proton users criticised the company's leadership on social media and explicitly called for Yen to resign; others said they were cancelling their paid plans. When Yen answered on Reddit, users downvoted his comment more than 1,200 times.

The ownership structure explains why an apology was the available remedy. Proton is owned by the Proton Foundation, whose board of trustees includes Yen alongside Tim Berners-Lee, Carissa Veliz, Dingchao Lu and Antonio Gambardella. One employee told 404 Media they would have preferred "real accountability from the leadership team" and said they felt little could happen because Yen holds the majority of Proton's shares, a claim the outlet attributes to the employee rather than to any corporate filing.

That same employee's reading of the meeting was that "the only reason we had this meeting is because he has a rebellion on his hands."

What to watch, and what cannot be watched

The checkable dates are the reorganisation next month and whatever the performance-review cycle does with the AI metrics that were supposed to be measured in an interview. Those are observable from outside, eventually. A communication failure is not.

If the re-interviews were the visible edge of a policy, cancelling them removes the visible edge and leaves the policy, which is roughly what the employees describe and roughly what the CEO said out loud. The pentest question from August is still unanswered in public, and it is the one a privacy company's paying users were actually asking. Nothing published this week suggests the answer has changed.