Proton AI Controversy: What 404 Media Reported

404 Media reported Proton's CEO pushed AI over staff security concerns. What Proton said, what its AI tools do with your data, and what is unconfirmed.

··6 min read

Bottom Line: 404 Media reported on 29 October 2026 that Proton’s CEO pushed staff to adopt AI and called a request for security testing “loser thinking”, and Proton did not dispute the reporting, according to 404 Media. We found nothing in the parts we could read that changes how Proton VPN handles user data. The claims about Proton’s internal conduct come from one outlet’s anonymous sources, not from our own confirmation.

Disclosure. vpn.com earns a commission when readers buy Proton VPN through our links, and from other VPN providers that compete with it. Proton VPN is ranked #3 of 22 on our Trust Score, at 85/100. This story does not change that score, and it contains no purchase links.

Status. Facts checked against our research record on 2 November 2026. We have not asked Proton for comment; its public statements are quoted as relayed by other outlets.

Proton's homepage: the headline "A better internet starts with privacy and freedom" above icons for Proton's apps, including Lumo AI.
Proton VPN logo Figure 1. Proton's homepage. Screenshot: proton.me, captured 3 Nov 2026.

What happened

Reports ran over five days, from 404 Media’s story on planned staff interviews to the cancellation of those interviews.

Timeline from late August to 30 October 2026: the CEO's pentest reply, 404 Media's reports on 26 and 29 October, and the re-interviews cancelled on 30 October.
Figure 2. How the story unfolded. Dates are the dates of the reports, except the late-August Slack exchange, which 404 Media dates but has not published. Source: 404 Media.

The graphic shows four steps. Two details sit outside it. The interviews were due 4 to 11 November, per 404 Media’s 26 October report[1] and Proton’s staff email as quoted by Tom’s Guide. And on 27 October Tom’s Guide published Proton’s public comment on the plan, relayed from a Proton comment on Reddit.

What 404 Media reported

On 29 October, 404 Media[2] published an investigation by Joseph Cox, based on eight current and one former Proton employee, Slack chats, meeting recordings and an AI manifesto written by the company’s chief technology officer. It reports that CEO Andy Yen is pushing Proton to build with AI despite employees’ security concerns.

Headline, byline and summary of 404 Media's 29 October 2026 investigation into Proton's AI push, by Joseph Cox.
Figure 3. The headline, byline and summary of 404 Media's investigation, cropped. Source: 404 Media, 29 October 2026; screenshot captured 3 Nov 2026.

According to the report, management set up an email address for staff to flag anything blocking AI adoption, and software tracks the share of AI-written code. It also quotes a Slack message from Yen saying 2026 will be the last year Proton hires engineers.

The sharpest exchange came in the security channel at the end of August. 404 Media says an employee asked for a few days to pentest new code before release, and that Yen answered with “loser thinking” and said it is “possible to be fast and secure”. He later told the team to “stop lovingly handcrafting all our code and use AI more.”

Employees told 404 Media they have security concerns about building security products with AI but feel unable to voice them, for fear that Yen will berate them in front of colleagues or threaten their jobs. The accounts are anonymous.

What Proton said and did

According to 404 Media, a Proton spokesperson did not dispute any of its reporting or its characterization of what is happening inside the company. The spokesperson said some people at Proton are unhappy with the company’s direction or misunderstand it, and that many are anxious.

Proton’s public line, as relayed by Tom’s Guide[3], is that AI is changing the industry and the company must evolve how it works. Proton also says it is not doing layoffs and using AI as an excuse, that it has a talent shortage and that it is hiring. That sits against the Slack message attributed to Yen about engineer hiring, and we could not resolve the difference.

On 30 October, after the all-hands meeting, Yen cancelled the staff re-interviews, according to 404 Media’s follow-up[4], which cites audio of the meeting. Goosed[5] reports, from employees it quotes, that the AI-usage criteria will instead be folded into regular performance reviews. One employee told 404 Media, “I don’t think he will change anything.” Goosed says staff reaction was split.

Some Proton users say publicly that they have cancelled or will not renew, and 404 Media reports social-media calls for Yen to resign. We have not read the original posts and found no published count of cancellations or revenue impact.

Reported versus confirmed

Almost everything in this story traces to one reporter’s investigation. The graphic below separates what was reported from what Proton has acknowledged.

Two columns: what 404 Media reported about Proton's AI push, and what Proton said or did in response. Most items are not confirmed by a second source.
Figure 4. What 404 Media reported against what Proton said or did. Sources: 404 Media, Tom's Guide, Goosed.

Where each item stands: Proton’s own staff email, as Tom’s Guide quotes it, confirms that the re-interviews were planned. That they were cancelled is reported by 404 Media and Goosed, and the meeting audio is unpublished. That AI use will be folded into performance reviews comes from Goosed alone, citing employees, and Proton has made no statement we found. Proton’s “no layoffs” line and the Slack message about engineer hiring are both Proton-side statements, and we could not resolve the difference. For user cancellations there is no count and no Proton statement, so we treat them as unverified. The pentest remark and the staff fear of raising concerns remain single-source: the Slack messages are unpublished, the employees anonymous, and Proton acknowledged unhappiness but not the conduct alleged.

What it means for Proton users

The governance claims rest on a single investigation built largely on anonymous sources, so treat them as reported, not established. The practical question for a Proton customer is which products use AI and what Proton promises about them.

ProductOn recordIndependent check
Proton VPNFive annual no-logs audit reports by one firm, Securitum[6], from 2022 to 2026, each commissioned by Proton. They cover the VPN only.No source independent of Proton and Securitum describes the 2022 to 2025 report contents.
Lumo, Proton’s AI assistantProton states[7] Lumo runs open-source models on servers it controls, keeps no logs of prompts or replies, does not train on user chats, and encrypts prompts so only Lumo’s GPU servers can decrypt them.One independent hands-on write-up says the prompt must be readable on those servers to be answered. No independent audit found.
Scribe, the Proton Mail writing assistantProton states[8] it can run on your device or on Proton’s no-logs servers, and does not train on inbox data.No independent audit found. The statement dates from a July 2024 announcement.
Top of Proton's "Lumo privacy" support page, with a list of sections on models, no-logs policy, encryption and training.
Figure 5. The top of Proton's Lumo privacy page, the source of Proton's statements about Lumo. Screenshot: proton.me, captured 3 Nov 2026.

Read plainly, Proton’s description of Lumo means its servers can read a prompt while answering it, so Lumo is not end-to-end encrypted the way Proton Mail is. The one independent source calls that fair but true of any hosted AI assistant, not only Proton’s.

If you use Lumo or Scribe, treat those privacy terms as Proton’s promises, not verified guarantees. If you use only the VPN, the reporting we read does not describe any change to how it handles user data. For the product itself, see our Proton VPN review and the VPN research index.

What this does not change

The Trust Score rates eleven product and service criteria, and company governance is not one of them, so this report does not change Proton VPN’s score. Under our methodology a breach, a privacy incident or a new audit would trigger a review.

What we could not confirm

  • Most of the investigation. 404 Media’s main article is partly paywalled. We read the free portion, roughly the first third.
  • The underlying evidence. The Slack messages and the meeting audio have not been published, so we could not inspect them.
  • Proton’s side. We read Proton’s public comment only as relayed by Tom’s Guide, which carries Proton VPN affiliate deals on the same page, not the original post.
  • User numbers. We found no count of cancelled subscriptions.

References

  1. J. Cox. 2026. Proton To ‘Re-Interview’ Staff About AI To See Who Gets To Keep Their Job. 404 Media (October 26, 2026). Retrieved November 3, 2026 from https://www.404media.co/proton-to-re-interview-staff-about-ai-to-see-who-gets-to-keep-their-job/.Archived: Wayback Machine snapshot
  2. J. Cox. 2026. Manifestos, Fear, and ‘Loser Thinking’: Inside Proton’s AI Reckoning. 404 Media (October 29, 2026). Retrieved November 3, 2026 from https://www.404media.co/manifestos-fear-and-loser-thinking-inside-protons-ai-reckoning/.Archived: Wayback Machine snapshot
  3. A. Stevanović. 2026. Proton to re-interview staff on AI use – and jobs may be on the line. Tom's Guide (October 27, 2026). Retrieved November 3, 2026 from https://www.tomsguide.com/computing/vpns/proton-to-re-interview-staff-on-ai-use-and-jobs-may-be-on-the-line.
  4. J. Cox. 2026. Proton’s CEO Cancels AI ‘Re-Interviews’. 404 Media (October 30, 2026). Retrieved November 3, 2026 from https://www.404media.co/protons-ceo-cancels-ai-re-interviews/.Archived: Wayback Machine snapshot
  5. Marty. 2026. Proton’s AI controversy is a trust crisis, and trust is all Proton sells. Retrieved November 3, 2026 from https://goosed.ie/news/protons-ai-controversy-is-a-trust-crisis/.
  6. Securitum. 2025. Securitum Protonvpn Nologs 2025. Retrieved October 18, 2026 from https://securitum.com/public-reports/securitum-protonvpn-nologs-2025.pdf.Archived: Wayback Machine snapshot
  7. Proton. n.d. Lumo privacy. Retrieved November 3, 2026 from https://proton.me/support/lumo-privacy.
  8. E. Maguire. 2024. Introducing Proton Scribe, a private writing assistant. Proton (July 18, 2024). Retrieved November 3, 2026 from https://proton.me/blog/proton-scribe-writing-assistant.Archived: Wayback Machine snapshot

Frequently Asked Questions

Did the Proton AI controversy change how Proton VPN handles my data?

Not according to anything we could read. The Slack exchange 404 Media quotes concerns pentesting before release, not the VPN apps. The main investigation is partly paywalled, so unread detail could exist. Proton VPN’s published audits cover its no-logs policy, and its Trust Score is unchanged.

Is Proton’s Lumo AI end-to-end encrypted?

No, going by Proton’s own description. Proton says prompts are encrypted so only Lumo’s GPU servers can decrypt them, which means those servers read a prompt to answer it. Proton also says Lumo keeps no logs and does not train on chats. We found no independent audit of Lumo.

Has Proton’s AI been independently audited?

We found no independent security or privacy audit of Lumo or of Scribe, the Proton Mail writing assistant. The no-logs audits Proton has published cover Proton VPN only, and Proton commissioned each of them. Privacy terms for its AI tools are therefore Proton’s own statements.

Should I cancel Proton because of this report?

That is your decision, and nothing in this report changes Proton VPN’s Trust Score. Check which Proton products you use: the reporting concerns company management and AI use, not VPN data handling that we could find. We found no published count of cancellations or revenue impact.