Here's what it can't read.
At two in the morning on April 17, 2026, the US House of Representatives held an emergency vote on a surveillance law that most Americans have never heard of. They failed to pass it. Twenty Republicans broke from their own leadership. Most Democrats opposed the measure. After hours of back-and-forth, Congress settled for a ten-day extension and kicked the fight down the road to April 30.
The law is Section 702 of the Foreign Intelligence Surveillance Act. The short version: it lets the NSA, FBI, CIA, and other intelligence agencies pull the calls, texts, and emails of foreign nationals from US tech and telecom companies without getting a warrant. The longer version is messier. When those foreigners communicate with Americans, the Americans' messages get swept up too. The FBI then searches that pile of data for information about US citizens. That's the part that has lawmakers fighting.
There were 349,823 surveillance targets under Section 702 in 2025, according to the Office of the Director of National Intelligence. That number was 246,000 in 2022. Whatever direction you think this is heading, it's heading there quickly.
Section 702 doesn't work by the government intercepting your phone out of the air. It works by the government going to Google, AT&T, Verizon, Meta, and everyone else in that club, and compelling them to hand over data. The companies comply because the law requires it. The data lives on their servers, because you sent it through their infrastructure. When you text your spouse through your carrier, that message traverses a network operated by a company that can be legally required to produce it.
In August 2024, Department of Justice overseers discovered the FBI had been using a "filtering tool" to search Americans' communications without following the rules Congress had set. In March 2026, the FISA Court found the problem extends beyond the FBI — similar tools are in use across the intelligence community. The oversight system that was supposed to catch abuses didn't catch them. It relies on the executive branch to police itself.
None of this is a conspiracy theory. The Brennan Center for Justice publishes every detail. The Office of the Director of National Intelligence confirms the target counts. Members of Congress from both parties are on record saying the current system doesn't protect constitutional rights. The tool works because your communications move through centralized infrastructure that can be subpoenaed, filtered, and searched.
Now picture a different kind of message. You type it into an app on your phone. It goes over Bluetooth to a small radio in your pocket. The radio transmits it on the 915 MHz ISM band to another radio a mile away. That radio passes it to another one, and another one, until it reaches the device you meant it for. The message is encrypted end to end. There is no carrier. There is no server. There is no company that can be compelled to produce anything, because no company was ever involved.
A radio network built on cryptographic addressing and end-to-end encryption doesn't have the choke points that Section 702 was designed to exploit. There is no backbone to tap, because the traffic never touches the internet.
That's a Reticulum mesh. Or a Meshtastic network with a private channel. Or MeshCore with proper encryption enabled. The specifics vary, but the structural point is the same. There is no telecom giant to subpoena, because no telecom giant carried the packet.
It's a fair question. Signal, Wire, and similar end-to-end encrypted apps do protect message content — even when a company hands over data under a court order, they can only produce encrypted ciphertext. In that narrow sense, Signal is also outside Section 702's reach for content. But Signal still runs on infrastructure. It requires a phone number to register, which ties it to a carrier. It requires an internet connection, which passes through a provider. Metadata — who you messaged, when, how often — can still be compelled from Apple, Google, or the carrier itself even when message content cannot. And if the government decides to block or throttle Signal at the network level, they have the leverage to do it.
A properly configured mesh — no MQTT bridging, private channel keys enabled, no internet path in the route — has none of those dependencies. The metadata doesn't exist on any server because there is no server. That's a structurally different guarantee, not just a stronger version of the same one Signal offers.
In July 2025 the Electronic Frontier Foundation published a piece titled "Radio Hobbyists, Rejoice! Good News for LoRa & Mesh." Author Bill Budington wrote that these technologies "break down the over-reliance on traditional network hierarchies" and build "resistance to censorship, control and surveillance" into the network architecture itself. The EFF has spent 35 years litigating against government surveillance overreach. When they publish an enthusiastic piece about the technology you're running in your backyard, it is worth paying attention to why.
It would be easy to frame this as a paranoid argument. It isn't. Most people, most of the time, have nothing to hide from the federal government, and Section 702 is used to track legitimate foreign intelligence targets. That's the design intent. The problem is that once a surveillance system exists, it gets used for things beyond its original purpose.
Section 702 was used to surveil Black Lives Matter activists. It was used to surveil participants in the January 6 Capitol attack. It has been used on members of Congress, journalists, and political donors — all technically incidental to foreign targets.
The point isn't that you personally need to worry about the NSA reading your grocery list. The point is that the system exists, the system is expanding, and the system has been repeatedly shown to exceed its own rules. Whether you lean left, right, or neither, the architecture is the architecture. A government that can search communications without a warrant today can do it to whoever it wants tomorrow.
Mesh networks don't solve this problem for everyone. They don't replace your phone. They don't work over the distances that cell networks do. But a properly configured mesh — private channel encryption enabled, no MQTT bridge to the internet, no third-party server in the path — creates something that structurally cannot be captured by Section 702 or anything like it. A category of communication that is yours, full stop. No carrier in the middle. No server to compel. No company required to comply. The qualification matters: a Meshtastic node with MQTT bridging enabled is still touching a server, and that server can be subpoenaed. The architecture only holds when the architecture is complete.
Congress has until April 30, 2026 to figure out what they want Section 702 to look like for the next several years. The extension that just passed was a two-week holding pattern, not a resolution. Whatever Congress decides will shape the legal surveillance environment in the United States for the foreseeable future. It will happen whether or not the mesh community pays attention. The outcome is outside our control.
What is inside our control is the infrastructure we build.
Every node deployed correctly — private channel, no MQTT, no internet bridge — is one more point of communication that doesn't pass through a choke point. Every person we teach to configure a Meshtastic node, a MeshCore repeater, or a Reticulum RNode properly adds one more piece to a communications layer that is structurally different from the one Congress is arguing about this month. The value isn't symbolic. It's architectural.
The system we're building is independent of the system being legislated, by design. And the fight happening in the Capitol this week is the reason that design choice matters.
Guides, hardware recommendations, and free configuration documentation are all at nodestar.net.