Skip to content
Security & Trust Talking point

Chrome and Firefox's local network prompt reportedly checked permission too late

At Black Hat, Moriya Pedael of GeoEdge relayed a finding that clicking block did not stop connection timing.

W
WebPulse Newsroom
AI-assisted · 2 min read
Share on X LinkedIn
Chrome and Firefox's local network prompt reportedly checked permission too late
In brief
  • Moriya Pedael said a researcher found, almost a year before her talk, that Chrome and Firefox connected to local devices before checking permission, so blocking did not stop timing.
  • The excerpts do not say whether browser makers have changed this order since. A prompt protects people only if it comes before the connection.

Moriya Pedael, a security researcher at GeoEdge, said that the local network access feature in Chrome and Firefox made its connection before it checked the user's permission. She credited the finding to a researcher who raised it almost a year before her talk. She spoke on the Black Hat show, in a Black Hat USA 2026 session about LANJack, an ad campaign.

What was said

LANJack was delivered through a mainstream ad. Pedael described how it tried to reach devices on the viewer's own network. Local network access is meant to stop a web page from touching internal devices unless the user gives permission.

She asked the room who thought most people would click allow without thinking. She called the response "a lot" of hands. Her comment: "So much for believing in humanity."

She then said it does not matter. According to the researcher she cited, the browser makes the connection to the devices before it checks for permission. So even if the user clicks block, the connection still happens and the attacker can still measure how long it takes.

That timing is useful to an attacker. Pedael explained that the script first looks for the router, because the fastest connection is the gateway. Then it tries addresses and ports on the local network. Any response means a device is there. She called it a little ironic that "a 404 error actually means that there is a device behind this".

Why it matters

Our reading: a permission prompt only counts as a control if nothing sensitive happens before the answer. On the account Pedael gave, this prompt had a bigger problem than the audience's doubts about users. Clicking block did not stop the first step.

For people who buy or run software, that changes the question. Do not stop at whether a product asks for consent. Ask what it does before the user answers, and whether the person's choice changes anything. A review that counts a dialog as a safeguard may be counting something that does not work.

The other side

The poll was a show of hands in one room, not a measured study of user behaviour.

The block finding was secondhand. Pedael relayed it from another researcher. The excerpts do not say that LANJack was seen probing networks after users clicked block.

The timing leak shows only that a device exists. Reading data from it needs further steps, which she tied to a technique called DNS rebinding.

The excerpts also leave the current state unknown. They do not say whether Chrome or Firefox has changed the order of the check since the finding. Nor do they cover other browsers.

Written by the WebPulse Newsroom with AI assistance, and checked by our editorial review: every quotation was verified against the recording's transcript. How we use AI.

The conversation this talking point comes from