5 million Swedish kronor, about $514,000. That is the size of the personal political donation that Mullvad VPN co-founder and co-owner Daniel Berntsson made to the Örebro Party, a populist Swedish regional party, in 2025. According to TechRadar, citing the Swedish outlet Flamman, the sum accounted for 72% of the party’s total income that year. One donor, nearly three quarters of a political party’s budget. The disclosure triggered subscription cancellations at Mullvad and, this week, a public reckoning at the Tor Project, Mullvad’s closest technical partner.
In a statement published October 10, the Tor Project said it is not terminating its relationship with Mullvad but is “setting clear boundaries.” The concrete changes are two: Tor has paused proactive co-branding with Mullvad, meaning joint promotional activity stops while the engineering work continues, and it has rewritten how Mullvad Browser is described on Tor’s own channels, removing language that suggested the two organizations share values. The technical collaboration itself continues under what Tor calls a narrowly scoped agreement.
What the collaboration actually is
Mullvad Browser is a Firefox-derived browser built by Mullvad together with the Tor Project: Tor Browser’s codebase minus the Tor network, meant to be run over a VPN instead. Tor’s statement lists what the partnership has produced for its side of the fork: improvements to the “structure, maintainability, and auditability” of Tor Browser’s codebase, feedback from users outside Tor Browser’s core audience, and a lower-risk environment for testing features that may later ship in Tor Browser itself. A feature that breaks Mullvad Browser breaks it without deanonymizing anyone; the same bug in Tor Browser carries a different cost. That is the engineering argument for keeping the relationship.
Tor also concedes the cost side of the ledger, which is unusual for a statement of this kind. The collaboration benefits Mullvad as well, and “the potential benefits to Mullvad of such a collaboration are difficult for some in our community to accept.” Continuing, Tor says, “carries a cost to trust for some in our community.”
How the decision was made
The process Tor describes reads like a risk assessment rather than a values debate: extensive discussions with community members, staff, and the board, a staff survey, and “detailed financial analysis and scenario planning.” The financial analysis matters because Tor’s funding is narrow. The Project does not disclose in the statement what Mullvad contributes, but it frames the privacy and anti-censorship field as “severely under-resourced” and its output as work that “requires and benefits from collaboration across organizations.” Scenario planning implies Tor modeled the case where it walks away from Mullvad and found the result worse for its users than the case where it stays and absorbs the reputational hit.
Tor’s statement never names Berntsson, the Örebro Party, or the 5 million SEK figure. It refers only to “a political donation by a Mullvad co-founder.” What it does say, in the closest thing to a position on the donation itself, is that “not all speech is equally compatible with our mission” and that Tor “strongly oppose[s] rhetoric that threatens other human rights and freedoms.” The boundary Tor draws is not between good and bad partners but between speech it defends as a right and speech it will not be seen to endorse.
Precedent and incentives
This is the standard free-software funding problem with a political twist. Anti-censorship tooling has few payers: the user base includes activists, journalists, and people under repression, most of whom cannot or should not pay in a traceable way. Corporate partners bring money, developers, and distribution, and each one brings its own risk of becoming the story. Tor’s answer, as written, is a two-tier model: shared engineering, which it treats as separable from the partner’s politics, and shared branding, which it treats as endorsement and can therefore pause. Whether that separation holds is a factual question about how users and donors actually behave. Tor’s own statement acknowledges it may not: “We know this decision will not satisfy everyone.”
The numbers to watch from here are the ones neither organization has published: what Mullvad contributes to the collaboration in money or engineering hours, how much of Tor Browser’s recent maintenance work runs through the shared codebase, and whether the fall in Mullvad subscriptions that followed the donation’s disclosure changes Mullvad’s capacity to fund its side. If the collaboration survives on Tor’s new terms while Mullvad’s revenue falls, Tor will have made the opposite of the usual trade: keeping the engineering dependency while cutting the marketing that made it tolerable to Mullvad.