Hard-coded RSA keys found in Watchfire controller firmware
Plaintext keys in patch binaries means the compromise surface extends to anyone who's ever touched Watchfire's filestore, this isn't a bug, it's a supply-chain trust failure.
TL;DR
CISA disclosed CVE-2026-5846, a hard-coded cryptographic key vulnerability in Watchfire Controller firmware affecting BC550, BC750, BC760, and BC760DC models deployed across defense, critical manufacturing, healthcare, and financial services sectors. The RSA private keys and corresponding X.509 certificates used for TLS on the controllers' web management interfaces were embedded in plaintext in patch files served from Watchfire's own filestore. Watchfire says it has patched all managed controllers and published firmware updates. The question is whether every unmanaged installation has applied them, and whether anyone else already extracted those keys.
This is the kind of advisory where the vulnerability class tells you more than the CVSS score. CVE-2026-5846 is a hard-coded cryptographic key (CWE-321) not a buffer overflow or an auth bypass. The RSA private keys and X.509 certs that authenticate TLS connections to Watchfire's BC-series controllers were baked into the firmware binaries, then shipped in plaintext inside patch files available from Watchfire's Remote Support filestore. Anyone who had access to those files (internally or externally) had the keys.
That's the operational reality the CVSS 5.7 doesn't capture. The vector assumes high attack complexity and high privileges. But if you already possess the private key from a patch binary, you're not exploiting a vulnerability in the traditional sense. You're a trusted node. You can stand up a TLS session with any affected controller, deliver signed firmware, and own the device. No privilege escalation required because the key is the privilege.
Watchfire's remediation is straightforward. The company patched all controllers under its management remotely. For unmanaged installations, the upgrade path is: BC550 to 12.31 SP1, BC750 to 11.34 or 12.36 SP1, BC760 to 12.41 SP1 or 14.00 SP1, BC760DC to 12.41 SP1. The patch disables the compromised certificate.
But the plaintext-in-filestore detail changes the risk calculus. A hard-coded key is bad. A hard-coded key that lived unencrypted in a distributable patch binary is worse, it means the exposure window isn't bounded by the device's deployment lifetime. It's bounded by however long those patch files sat in that filestore and whoever had access to them. CISA notes no known public exploitation as of July 30, but the acknowledgment credits researcher James Tillson, meaning the keys were extractable by anyone who looked.
For defense contractors and critical-infrastructure operators running these controllers, the Monday-morning question isn't just "did we patch." It's "did we rotate every credential and re-architect the trust boundary around these devices," because the old keys aren't just compromised, they were never secret.
Published ·Deep Fathom