CVE-2023-25725 [Critical] | Security Bypass in Haproxy

HAProxy before 2.7.3 may allow a bypass of access control because HTTP/1 headers are inadvertently lost in some situations, aka "request smuggling." The HTTP header parsers in HAProxy may accept empty header field names, which could be used to truncate the list of HTTP headers and thus make some headers disappear after being parsed and processed for HTTP/1.0 and HTTP/1.1. For HTTP/2 and HTTP/3, the impact is limited because the headers disappear before being parsed and processed, as if they had not been sent by the client. The fixed versions are 2.7.3, 2.6.9, 2.5.12, 2.4.22, 2.2.29, and 2.0.31.

#	Date	Old EPSS score	New EPSS score	Delta (New - Old)
1	2026-06-15	17.54%	5.49%	-12.04%
2	2026-05-13	20.04%	17.54%	-2.50%
3	2026-04-18	—	20.04%	—

Date

Old EPSS score

New EPSS score

Delta (New - Old)

2026-06-15

17.54%

5.49%

-12.04%

2026-05-13

20.04%

17.54%

-2.50%

2026-04-18

—

20.04%

—

Base score	Version	Severity	Vector	Exploitability	Impact	Score source
9.1	3.1	CRITICAL	`CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:H/A:H` Click to expand Attack vector (AV:N) Could be attacked over the internet or any normal routed network—not just someone sitting at the machine. Attack complexity (AC:L) Once they can reach the bug, pulling it off is straightforward—no weird race conditions or rare setup. Privileges required (PR:N) No account or special rights needed—anonymous or random user is enough. User interaction (UI:N) Nobody has to click “OK” or open a trap file; it can work without a victim helping. Scope (S:U) Damage stays in the same “trust bubble” as the broken component—no big spill into unrelated systems. Confidentiality (C:N) Doesn’t really leak secrets in a meaningful way. Integrity (I:H) They could widely tamper with or forge data—trust in the data is badly hurt. Availability (A:H) Could take the service down hard or make it unusable for people who depend on it.	3.9	5.2	[email protected]
9.1	3.1	CRITICAL	`CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:H/A:H` Click to expand Attack vector (AV:N) Could be attacked over the internet or any normal routed network—not just someone sitting at the machine. Attack complexity (AC:L) Once they can reach the bug, pulling it off is straightforward—no weird race conditions or rare setup. Privileges required (PR:N) No account or special rights needed—anonymous or random user is enough. User interaction (UI:N) Nobody has to click “OK” or open a trap file; it can work without a victim helping. Scope (S:U) Damage stays in the same “trust bubble” as the broken component—no big spill into unrelated systems. Confidentiality (C:N) Doesn’t really leak secrets in a meaningful way. Integrity (I:H) They could widely tamper with or forge data—trust in the data is badly hurt. Availability (A:H) Could take the service down hard or make it unusable for people who depend on it.	3.9	5.2	134c704f-9b21-4f2e-91b3-4a467353bcc0

Base score

Version

Severity

Vector

Exploitability

Impact

Score source

9.1

3.1

CRITICAL

CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:H/A:H Click to expand

Attack vector (AV:N): Could be attacked over the internet or any normal routed network—not just someone sitting at the machine.
Attack complexity (AC:L): Once they can reach the bug, pulling it off is straightforward—no weird race conditions or rare setup.
Privileges required (PR:N): No account or special rights needed—anonymous or random user is enough.
User interaction (UI:N): Nobody has to click “OK” or open a trap file; it can work without a victim helping.
Scope (S:U): Damage stays in the same “trust bubble” as the broken component—no big spill into unrelated systems.
Confidentiality (C:N): Doesn’t really leak secrets in a meaningful way.
Integrity (I:H): They could widely tamper with or forge data—trust in the data is badly hurt.
Availability (A:H): Could take the service down hard or make it unusable for people who depend on it.

3.9

5.2

[email protected]

9.1

3.1

CRITICAL

CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:H/A:H Click to expand

Attack vector (AV:N): Could be attacked over the internet or any normal routed network—not just someone sitting at the machine.
Attack complexity (AC:L): Once they can reach the bug, pulling it off is straightforward—no weird race conditions or rare setup.
Privileges required (PR:N): No account or special rights needed—anonymous or random user is enough.
User interaction (UI:N): Nobody has to click “OK” or open a trap file; it can work without a victim helping.
Scope (S:U): Damage stays in the same “trust bubble” as the broken component—no big spill into unrelated systems.
Confidentiality (C:N): Doesn’t really leak secrets in a meaningful way.
Integrity (I:H): They could widely tamper with or forge data—trust in the data is badly hurt.
Availability (A:H): Could take the service down hard or make it unusable for people who depend on it.

3.9

5.2

134c704f-9b21-4f2e-91b3-4a467353bcc0

OS Trackers for CVE-2023-25725

vendor	priority	summary	link
`alpine`	—	CVE-2023-25725: 1 source package rows (haproxy); 5 state rows across 5 repos (3.19-main, 3.20-main, 3.21-main, 3.22-main, edge-main); fixed 0, open 5.	https://security.alpinelinux.org/vuln/CVE-2023-25725
`debian`	not yet assigned	CVE-2023-25725 not yet assigned priority: Debian including 1 source packages (haproxy), 5 status rows across 5 suites (bookworm, bullseye, forky, sid, trixie): resolved 5.	https://security-tracker.debian.org/tracker/CVE-2023-25725
`redhat`	high	—	https://access.redhat.com/security/cve/CVE-2023-25725
`suse`	critical	CVE-2023-25725 severity critical: SUSE including 9 source package names (2.0.31.3.5.391:haproxy-2.0.31-150200.11.20.1, haproxy, …), 22 product×package rows across 22 product lines (Container ses/7.1/ceph/haproxy, HPE Helion OpenStack 8, … (22 product lines)): Fixed 13, Known Not Affected 9.	https://www.suse.com/security/cve/CVE-2023-25725/
`ubuntu`	medium	CVE-2023-25725 medium priority: Ubuntu including 1 source packages (haproxy), 11 status rows across 11 suites (bionic, focal, jammy, kinetic, lunar, mantic, noble, oracular, trusty, upstream, xenial): released 9, ignored 1, needs-triage 1.	https://ubuntu.com/security/CVE-2023-25725

Affected software / configurations for CVE-2023-25725

Vendor	Product	Version	Raw CPE
haproxy	haproxy	< 2.0.31	cpe:2.3:a:haproxy:haproxy::::::::
haproxy	haproxy	>= 2.1.0, < 2.2.29	cpe:2.3:a:haproxy:haproxy::::::::
haproxy	haproxy	>= 2.3.0, < 2.4.22	cpe:2.3:a:haproxy:haproxy::::::::
haproxy	haproxy	>= 2.5.0, < 2.5.12	cpe:2.3:a:haproxy:haproxy::::::::
haproxy	haproxy	>= 2.6.0, < 2.6.9	cpe:2.3:a:haproxy:haproxy::::::::
haproxy	haproxy	>= 2.7.0, < 2.7.3	cpe:2.3:a:haproxy:haproxy::::::::
debian	debian_linux	10.0	cpe:2.3:o:debian:debian_linux:10.0:::::::*
debian	debian_linux	11.0	cpe:2.3:o:debian:debian_linux:11.0:::::::*

References for CVE-2023-25725

URL	Tags
https://git.haproxy.org/?p=haproxy-2.7.git%3Ba=commit%3Bh=a0e561ad7f29ed50c473f5a9da664267b60d1112
https://lists.debian.org/debian-lts-announce/2023/02/msg00012.html	Mailing List Third Party Advisory
https://lists.fedoraproject.org/archives/list/package-announce%40lists.fedoraproject.org/message/FPTJQHKUEU2PQ7RWFUYAFLAD4STEIKHU/
https://lists.fedoraproject.org/archives/list/package-announce%40lists.fedoraproject.org/message/JM5NCIBTHYDTLPY2UNC4HO2VAHHE6CJG/
https://www.debian.org/security/2023/dsa-5348	Third Party Advisory
https://www.haproxy.org/	Product

URL

CVE-2023-25725

Exploit prediction scoring system (EPSS) score for CVE-2023-25725

Common vulnerability scoring system (CVSS) metrics for CVE-2023-25725

Weakness enumeration for CVE-2023-25725

OS Trackers for CVE-2023-25725

Affected software / configurations for CVE-2023-25725

References for CVE-2023-25725