Futureweb

Privacy

Proton is hosting a map of Flock cameras and says the underlying inventory is a December 2025 snapshot. Proton says it does not host the dataset itself.

Proton's map counts more than 170,000 cameras and over 130,000 other devices, from a December 2025 snapshot. Proton says it does not host the dataset.

The short version

A December 2025 snapshot, not a live census. Proton says the inventory counts more than 170,000 cameras and over 130,000 other devices, and that Flock has publicly cited around 120,000 cameras, so Proton calls the camera count about 40 percent higher. Proton says it hosts the map and not the underlying data. This page does not describe how the inventory was obtained.

A snapshot, hosted as a map

Proton's Flock camera map page says the inventory behind the map includes more than 170,000 cameras and over 130,000 other devices. It says the snapshot is from December 2025, not a feed of cameras switching on this morning. It says Proton hosts the map and does not host the underlying dataset. The company's October 23 writeup on its own site makes the same distinction. Tom's Hardware's October 24 story is the outside account on the wire, and its headline rounds the holdings past 300,000. Adding Proton's two figures gets a reader to that neighborhood. This page will keep the split, because a camera and a different device are not the same object.the Gmail policy line is a different privacy change.

Forty percent is Proton's comparison

Proton says Flock has publicly cited around 120,000 cameras, and that the inventory's camera count is about 40 percent above that citation. The 40 percent is Proton's arithmetic against a number it attributes to Flock, not a number Flock published on Proton's page. If Flock's public count has moved since the citation Proton used, the gap moves with it. A December 2025 snapshot cannot see a camera installed in October 2026, and it cannot see one removed. The map's honesty is the date in the corner.a session is not a camera.

The token is provenance, not a procedure

Proton says the inventory comes from work by Joshua Michael, who in November 2025 reported an exposed access token to Flock, and who published the inventory in December 2025. That is the chain of custody the company is willing to put on the page: a report, a publication, a snapshot. This page does not describe how a token was reached, what was requested, or where a reader would ask again. Tom's account of a cease-and-desist is Tom's legal color. It is not a how-to, and it is not this page's. DeFlock, a crowdsourced map, is a different project and not these counts.the key stays with the holder.

Two of the sources are Proton

The map page and the Substack post are the same company. Tom's Hardware is the outside account. Using two Proton links is how a reader checks that the blog and the map agree on the snapshot, not how this page pretends a second lab counted cameras. Where Tom's says the earlier public map was United States only and Proton's view is wider, that geographic claim stays Tom's. This page does not audit the pins. Asked directly, Why do 170,000 and 300,000 both appear? Proton's page splits cameras and other devices. Tom's headline adds them toward 300,000. This page keeps the split, and treats the round total as the headline's sum.what a TLS certificate binds is a different object.

A Geneva line is an address, not a finding

Proton's corporate address is in Geneva. Tom's mentions Swiss hosting. Neither fact is an indicator of compromise, a server name, or a location of a camera. A company that says it hosts a map from Switzerland is describing its own office and its own hosting choice. Readers who came for a list of devices will not find one here. The count is the news. The coordinates of individual cameras stay on the map Proton operates, under whatever terms that page sets.

Who is in scope

People who want to know what the map claims to be: a December 2025 inventory, larger than Flock's public camera citation on Proton's telling, split between cameras and other devices, hosted as a visualization rather than as the raw data. City governments, drivers, and Flock itself are the subjects of the map. They are not, from this page, given a query. Asked directly, Does Proton host the camera dataset? The page says Proton hosts the map and does not host the underlying dataset. Those are different roles.

What the page will not resolve

It will not decide whether Flock's 120,000 or Proton's 170,000 is the truer census, because they are not counts of the same day, and only one of them is sitting on a page this brief treats as primary. It will not treat Tom's 300,000 headline as a third measurement. It will not describe retrieval. A privacy story about a camera network can state the size of a published snapshot without becoming a client for the network.the advisory pages are a different desk.

October 23 and October 24

Proton's post is October 23. Tom's is October 24. The snapshot is December 2025. Three dates, three meanings. The newest date is a news story. The oldest date is the data. Confusing them is how a map starts to look live. Asked directly, Is the Proton map a live count of Flock cameras? Proton says no. It is a snapshot from December 2025. A map that loads today can still be showing last December.

The limit

No retrieval steps. No camera list. No claim the map is current. The record is Proton's split count, the December 2025 label, and the statement that Proton does not host the dataset. The short answer this page will stand on is this: A December 2025 snapshot, not a live census. Proton says the inventory counts more than 170,000 cameras and over 130,000 other devices, and that Flock has publicly cited around 120,000 cameras, so Proton calls the camera count about 40 percent higher. Proton says it hosts the map and not the underlying data. This page does not describe how the inventory was obtained.

Sources

The reports this brief is filing. Futureweb did not republish them.

  1. Proton
  2. Proton blog
  3. Tom's Hardware

Questions

Is the Proton map a live count of Flock cameras?

Proton says no. It is a snapshot from December 2025. A map that loads today can still be showing last December.

Does Proton host the camera dataset?

The page says Proton hosts the map and does not host the underlying dataset. Those are different roles.

Why do 170,000 and 300,000 both appear?

Proton's page splits cameras and other devices. Tom's headline adds them toward 300,000. This page keeps the split, and treats the round total as the headline's sum.

What does Proton say about Flock's own number?

Proton says Flock has publicly cited around 120,000 cameras, and that the inventory's camera count is about 40 percent higher. The comparison is Proton's.

Does this page explain how the inventory was gathered?

No. Proton says a researcher reported an exposed token to Flock in November 2025. The method stays off this page. The snapshot date is the fact a reader needs.