David Buchanan (retr0id) argues the C2PA content-provenance ecosystem faces its own eMudhra-style trust failure, showing he can get a Google intermediate to sign arbitrary leaf keys and explaining why that breaks the C2PA trust model.
C2PA's eMudhra Moment
By David Buchanan (aka retr0id), 29th October 2026
The eMudhra incident has shown how the modern Web PKI ecosystem reacts to a CA compromise. All things considered, I think it went smoothly—all the web browsers I use picked up the revocations swiftly. But it has also been a clusterfuck, with eMudhra trying to downplay their responsibilities and refusing to elaborate on how exactly the failure occurred. There's also a long tail of TLS clients that don't have a great way to handle revocations, or are otherwise not getting their trust stores updated in a timely manner.
The C2PA ecosystem tries to solve a very similar problem to Web PKI, but it is comparatively in its infancy. The media verification tools you'll find in search result top-spots today don't support revocation-checking at all.
So how would the C2PA ecosystem react to a similar compromise? Let's find out!
The above demonstrates a S/MIME encrypted message, with a signature that chains to Google's C2PA root CA, which is in the official C2PA trust list.
The vulnerabilities I exploited to produce this chain are not eligible for bug bounty rewards, and are in fact already publicly known. But in the spirit of our little experiment I'm not going to elaborate.
The astute observer will note that my demo is "not the same" as the eMudhra compromise, because it's merely a leaf cert that has a private key of 4, rather than a trusted intermediate.
I can get Google's intermediate to sign any leaf key I like (via an undisclosed method), and that "4" key is just how I'm proving it.
In the context of TLS that's kinda how cert issuance works normally, but it completely breaks the C2PA trust model.