On June 26, the FBI published a public service announcement with a plain warning: Russian intelligence services are stealing Signal users' message histories, and they are doing it without touching the target's phone.
The mechanism is worth understanding because nothing about it is technical. The attacker sends a message posing as Signal support and claims a mandatory verification step is required. The target is walked through enabling Signal's backup feature and then asked to paste in the backup recovery key. That key lets the attacker restore the target's message history on a device the attacker controls. The FBI attributes the campaign to Russian intelligence services, tracked as UNC5792 and UNC4221, and names the targets: current and former government officials, military personnel, political figures, and journalists. The same day, Ukraine's Security Service described a coordinated campaign of support-impersonation messages aimed at officials and citizens across Ukraine, Europe, and the United States.
This was not the first warning. In February 2025, Google's threat intelligence team documented the same groups abusing Signal's linked-devices feature. A malicious QR code, disguised as a group invite, silently added an attacker-controlled device to the target's account. From that point, every message the target sent or received was delivered to the attacker in real time.
Here is the detail that matters for anyone who moved sensitive conversations onto a secure messaging app: in every one of these cases, Signal's encryption worked exactly as designed. Nobody broke it. Nobody needed to.
What the encryption actually covers
End-to-end encryption protects the message in transit. Between the moment a message leaves one device and the moment it arrives on another, nobody in the middle can read it. That guarantee is real, it is strong, and it is the reason Signal and WhatsApp were the right choice for sensitive communications in the first place.
But a conversation does not live only in transit. It lives in the message history, which a backup recovery key can restore anywhere. It lives on every device linked to the account, including one added by a QR code scanned in the wrong context. It lives behind the account itself, which can be taken through social engineering without any software vulnerability. And it happens in a physical room, which no app has ever claimed to secure.
Each of those is an attack path that leaves the encryption untouched. The campaigns above used the first two and the third. The fourth is the oldest one in the book, and it did not go away because the messages got encrypted.
Where the false assurance comes from
The gap is not carelessness. It is a category error that almost every organization makes: treating the secure app as the communications security program, rather than as one control inside it.
The pattern is familiar. A deal window opens, or a board matter turns sensitive, and the instruction goes out: take it to Signal. The instruction is sound. What rarely follows is the rest: who verifies the recovery settings on each participant's account, who audits the linked-devices list, what the procedure is when "support" reaches out, and what room the calls actually happen in. The app was assessed once, informally, by reputation. The posture around it was never assessed at all.
An attacker reads that gap the way the FBI advisory describes: do not attack the encryption, attack the person and the account layer around it. The morning-hours timing noted in the Ukrainian advisory is the tell. These campaigns are engineered around human attention, not software flaws.
What a communications security posture includes
The fix is not exotic and most of it costs attention rather than money.
Account hardening on every participating device. Registration lock enabled, backup settings deliberately chosen, an app-level screen lock in place, and the recovery key stored where no message thread can reach it.
A linked-devices audit on a schedule, not once. The list is short and takes a minute to read. An unfamiliar entry is a finding, not a curiosity.
A standing rule for support contact. No legitimate platform support asks for a recovery key, a PIN, or a verification code. Any message that does is treated as hostile and reported, not answered.
Device discipline for the meeting window itself. What is on the phone, what the phone can sense, and whether it belongs in the room at all during the conversations that matter most.
The room, assessed. For recurring sensitive discussions, a technical surveillance countermeasures sweep answers the question the app cannot: is this space clean enough for this conversation.
None of this argues against Signal. Signal remains the right transport for sensitive traffic, and the encryption is not the weak point. That is precisely the reason the attacks have moved to the layers around it.
When to do this work
The honest timing is the same as for most protective work: before the sensitive period, not during it. A deal window, contested litigation, a leadership transition, or a rising public profile all raise the value of the conversations and the attention paid to them. Verifying the account layer takes days. Building the habit of a linked-device audit takes one calendar entry. Sweeping the room where the quarterly discussions happen takes one visit. All of it is cheaper and quieter than working out, after the fact, how a confidential position reached the other side of the table.
Triangle Point Solutions runs communications security assessments for executives and the teams around them: the account and device layer, the meeting-window discipline, and the physical sweep of the rooms where sensitive conversations actually happen. Tell us your stack, and we can help audit it for deficiencies and improvements. If your organization moved its most sensitive traffic to a secure app and stopped there, a scoped review of what sits around that app is the logical next step.