ssl_verify_fun accepts certificates for the wrong hostname
A library that checks who a secure connection is really talking to can be fooled by a special certificate, letting an attacker in the middle pose as the real server. Normal setups are protected, but some are not.
| Severity | CriticalCVSS 9.1 of 10 |
|---|---|
| Being exploited? | UnclearNo reports either way yet |
| Fix available? | Yesssl_verify_fun 1.1.8 or later · since 30 Oct 2026 |
| How sure is this? | Official sourceBacked by an official source |
How the attack works simplified
- 1Attacker in the middleIntercepts the connection to the server
- 2Crafted certificateWildcard like f*.example.net
- 3Hostname check passesWildcard matches across dots
- 4Impersonation succeedsClient trusts the wrong peer
- ResultAttacker impersonates a TLS peer
What happened
In ssl_verify_fun (Erlang), the hostname matcher try_match_wildcard/4 lets a partial-label wildcard such as f*.example.net match across dots, and lets *.168.1.1 match the IP 192.168.1.1, violating RFC 6125.
Are you affected?
You are affected if you run ssl_verify_fun from 1.0.1 before 1.1.8 and use it with SNI disabled, verify client certificate hostnames with it, or call verify_cert_hostname/2 directly.
ssl_verify_fun from 1.0.1 before 1.1.8
What to do
- Upgrade ssl_verify_fun to version 1.1.8 or later
- Confirm that clients keep server_name_indication set, which blocks this mismatch in Erlang/OTP 20.0 and later
- Review any direct calls to verify_cert_hostname/2 and any server-side client certificate hostname checks
Timeline
- 30 Oct 2026, 21:16 UTCPublicly disclosed
- 30 Oct 2026, 21:16 UTCReported by NVD · NVD
- 30 Oct 2026, 21:20 UTCPicked up by Zero Radar
Sources 1 report, oldest first
Official vendor, government CERT or advisory database
- CVE-2026-117666NVD · 30 Oct 2026, 21:16 UTC
Technical details
- ID
- CVE-2026-117666
- CVSS
- 9.1 / 10
- Disclosed
- 30 Oct 2026, 21:16 UTC
- On Zero Radar
- 30 Oct 2026, 21:20 UTC (4 min after disclosure)
- Last change
- 30 Oct 2026, 21:20 UTC
The plain-language parts and the attack diagram were written by an AI model (anthropic/claude-haiku-5.5) from the sources above. Facts like versions, exploitation and dates come from the sources themselves. How Zero Radar works