VPN for Journalists Abroad: I Had Eighteen Minutes to Get the Interview Out

The newsroom could read my transcript, but it could not hear the woman who had risked meeting me. I was in a Kampala hotel, trying to send an audio interview, two photographs and my verification notes before Uganda’s election internet shutdown began. The secure upload had accepted the text, stalled on the recording and then discarded the entire transfer. I blamed the 186-megabyte audio file, divided it into smaller parts and watched the first one fail again.

My editor sent a message at 5:42 p.m.:

Connectivity may go at six. Send the reporting package now.

Two days before Uganda’s January 15, 2026 general election, authorities ordered internet access suspended from 6 p.m. CPJ had also documented assaults on journalists covering the campaign. (Committee to Protect Journalists)

I had eighteen minutes.

The interview was with a local election observer who had described pressure on volunteers documenting incidents outside polling stations. She had agreed to be identified by role and district, but not by name. The transcript was useful. Her voice, pauses and corrections were evidence my editor needed before publication.

The story could wait.

The verification could not.

The hotel internet looked normal until I used it for reporting

The hotel Wi-Fi loaded ordinary websites quickly.

News pages opened. Messages arrived. A speed test showed enough bandwidth to send the recording several times over.

That made the failed upload feel like a file problem.

It was not.

The network passed routine browser traffic, but the secure transfer repeatedly slowed to zero. My usual VPN connected for a few seconds, lost its route and reconnected. Each interruption forced the newsroom uploader to begin again.

Travelers encounter the same mismatch on hotel Wi-Fi: ordinary browsing works while a VPN connection becomes unstable or unusable. (Reddit)

For a tourist, that is an annoyance.

For a journalist with source material and a closing network window, it determines what reaches the newsroom.

My familiar provider offered more choices than I had time to test

I had used the same major VPN provider for years.

It had a long public history, a large support operation and servers across several nearby countries. Those were reasonable reasons to trust it on assignment.

I selected Kenya.

The VPN connected.

The newsroom’s upload page opened, and the first audio segment began moving.

At 38 percent, the tunnel dropped.

I tried South Africa.

That route connected more slowly but reached 61 percent before the hotel network interrupted it.

The app suggested an automatic server.

I accepted.

This time, the uploader rejected the transfer before it began because the session had changed.

The provider could open international websites and protect short browsing sessions. But each new route gave the newsroom another connection to evaluate and another incomplete file to discard.

At 5:47, I still had no audio in the newsroom.

The server list had become a clock.

Turning off the VPN made the wrong thing faster

I disconnected the VPN and tried the upload directly.

The first audio segment reached 72 percent in less than a minute.

For a moment, I considered letting it finish.

Then I looked at the folder waiting behind it.

The original recording.

A photograph of the interview location.

A note containing the source’s agreed description.

A verification document with phone numbers I had not yet removed.

Our newsroom travel protocol treated a VPN as one layer of protection for reporting traffic on public networks. Journalist security guidance makes the same practical recommendation. (Freedom of the Press Foundation)

The direct upload was faster.

It was also the route I did not want carrying unpublished source material through a hotel network.

I stopped it.

That cost another minute, but it clarified the decision. I was not looking for any connection that moved data. I needed a protected route that could remain usable until the complete reporting package arrived.

The browser extension reached the newsroom but not the evidence

A colleague had previously recommended a reputable free browser extension for emergencies.

I installed it and selected a European route.

The newsroom website opened.

The article draft saved.

Messages to my editor went through.

For thirty seconds, it looked like the simplest answer.

Then I opened the encrypted transfer application we used for larger files.

It remained disconnected.

The extension covered the browser. It did not cover the desktop transfer tool, the metadata utility or the messaging app where my editor was checking each arrival.

I could paste the story into the content system.

I could not send the material that supported it.

That ended the experiment.

A reporting workflow does not remain inside one tab. The notes, recordings, photographs and newsroom tools have to travel through the same protected connection.

At 5:50, ten minutes remained.

The smaller app treated the network as the problem

I had installed OnlydogVPN^↗ before the trip but had left it as a backup.

The established provider had more countries, more ratings and a much longer public record. The smaller service had fewer locations and less independent history.

Those differences had influenced my choice at home.

They mattered less while the hotel network repeatedly disrupted the connection I already trusted.

I selected the restrictive-network preset and connected the full laptop.

The newsroom uploader reopened.

I started with the original audio file rather than the smaller pieces.

The progress bar passed 38 percent.

Then 61.

The hotel Wi-Fi hesitated.

The transfer slowed but did not disappear.

At 82 percent, the Wi-Fi dropped completely.

I turned on my phone’s roaming hotspot.

The laptop changed networks.

The VPN recovered, and the upload continued from where it had paused.

At 5:56, the newsroom displayed:

Audio received

I sent the two photographs.

Then the transcript.

Then the verification note after removing the source’s direct number.

My editor replied at 5:57:

Everything received. Voice matches transcript. Publishing.

That completed the task.

The source’s account was no longer trapped on my laptop. The newsroom had the material needed to verify, edit and publish it.

The connection stayed out of the reporting

The service combines traffic obfuscation with an HTTP/3-based connection designed for restrictive and changing networks.

The practical result was simple.

The established provider repeatedly connected but could not keep the transfer alive.

The browser extension protected only the visible part of the workflow.

The backup carried the entire laptop and recovered when the connection moved from hotel Wi-Fi to the phone hotspot.

I could not observe the hotel’s or internet provider’s internal filtering rules. I could compare what reached the newsroom.

For this assignment, completing one protected transfer mattered more than having a long list of nearby servers.

A VPN could not defeat a complete shutdown

At 5:59, my editor asked for one final detail:

Was the street outside the interview location open when you left?

I checked my notes and replied:

Open, with two police vehicles at the junction.

The message sent.

Then the newsroom chat stopped updating.

The hotel Wi-Fi still showed connected, but international pages no longer loaded. My roaming phone also lost usable data.

A VPN could not create an international route after the national connection had been cut. Uganda’s election shutdown lasted five days, and the initial partial restoration still left several major communication platforms blocked. (Access Now)

That was why the final eighteen minutes mattered.

The backup had kept the remaining route usable until the interview, photographs and verification notes were safely outside the country.

The published article created one smaller problem

The editor managed to publish the first version from outside Uganda.

The article used the source description we had agreed on and did not include her name.

I still needed to review the mobile proof for an accidental caption or identifying detail.

The hotel internet was gone, but the proof had already downloaded to my phone before the cutoff.

I opened it offline.

The caption identified only the district and date.

The photograph did not show the source.

I sent my editor a message through the hotel desk’s international phone service:

Proof checked. No identifying detail.

That final review did not depend on the VPN.

It depended on the reporting package arriving early enough for the newsroom to prepare the proof before the shutdown.

The successful transfer had created time for editorial caution instead of forcing us to choose between publishing quickly and protecting the person who spoke to me.

The upload speed was never the real standard

Before that assignment, I compared travel VPNs by speed, country count and whether they connected on the first attempt.

Those tests were too easy.

The Kampala hotel had fast Wi-Fi.

The major provider connected several times.

The free extension opened the newsroom immediately.

None of those results delivered the interview.

For a traveling journalist, the useful test is the complete reporting path:

Can the connection establish itself on a network that disrupts familiar VPN traffic?

Does it protect the whole device rather than one browser tab?

Can the file transfer continue when the hotel connection weakens?

Can it recover when the laptop moves to mobile data?

Does the editor receive the recording, notes and verification before the reporting window closes?

The smaller service had fewer locations and a shorter public history than the familiar provider.

It was also the connection that turned eighteen remaining minutes into a verified story outside the country.

At 6 p.m., the hotel still showed full Wi-Fi bars and no route to my editor. The interview mattered because its last file had crossed the border three minutes earlier.