Inside Proton’s push to make AI write more of its code

Proton is pushing staff to use AI more widely, with interviews about employees’ AI use potentially affecting their roles and continued employment, 404 Media reports. The account also describes AI writing 90 per cent of the company’s code in October, raising a pointed question for a privacy-focused technology firm: how quickly can AI adoption move without leaving security and software quality behind?

Watch Desk analysis

What happened

404 Media says it spoke to eight current and one former Proton employee and reviewed internal Slack messages, meeting recordings and a company AI manifesto. Its report describes a sharp change from an initial ban on workplace AI to approval for uses including processing financial data, drafting customer-facing responses and writing code, with exceptions for some sensitive or research-related code.

Proton CTO Bart Butler reportedly said in a meeting that AI wrote 90 per cent of the company’s code in October. The report says much of it was written using Anthropic’s Claude. It also describes plans to re-interview some staff about their AI use, with the results potentially affecting roles and continued employment.

Proton’s spokesperson did not dispute the report’s account of the company’s direction, 404 Media says. The spokesperson acknowledged that some employees were unhappy or anxious, and said the company had at times been too blunt or too slow to explain its decisions.

Why it matters

This is not just another company telling staff to try a chatbot. Proton sells privacy and security-focused services, while the reported changes reach into code-writing and security work. The central tension is practical: using AI to move faster may change how software is built, but speed does not by itself settle questions about review, security or who is accountable when something goes wrong.

The report also describes a consequential workplace policy, not simply a tooling preference. If AI fluency becomes part of how employees are assessed, the details of the interviews and their effect on jobs matter as much as the adoption figures.

Our read

The report offers a rare look at how an AI strategy can feel from inside a company whose customers value trust and security. Proton’s stated response recognises that staff are anxious and that leadership has not always explained itself well. That is a useful acknowledgement, but it leaves the harder questions on the table: what checks accompany AI-written code, and how will employees be judged fairly?

What to watch

• What Proton says about the scope and consequences of its staff interviews.
• How the company reviews AI-written code, particularly in security-sensitive products.
• Whether Proton provides more detail on its AI-use rules and the exceptions for sensitive work.

Discussion spark: Should a company be able to make AI adoption a factor in decisions about employees’ jobs, or should it first have to show that the tools improve work without weakening security and quality?

Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.