VulnSocial

CVE-2026-32882General

MEDIUMCVSS 7.1 · HIGH

Exploitation ongoing with high activity in latest observed window (1 mentions)

Immediate actions

  • Patch affected systems immediately
  • Assume compromise if assets are exposed

Recommended action window: Immediate (within 24h)

NVD description

libheif is a HEIF and AVIF file format decoder and encoder. Versions 1.21.2 and prior contain a heap buffer over-read in HeifPixelImage::overlay() in libheif/pixelimage.cc. When compositing an overlay image (iovl) whose child image has a different bit depth for the alpha channel than for the color channels, the function indexes into the alpha plane using the color channel stride (in_stride) instead of the previously retrieved alpha_stride, causing reads past the end of the alpha buffer (up to 3,123 bytes for a 100×50 image with 10-bit color and 8-bit alpha). A crafted HEIF file can exploit this to cause a denial of service (crash) or potentially disclose adjacent heap memory through leaked bytes embedded in the decoded output pixels. This issue has been fixed in versionThis issue has been fixed in version 1.22.0.

Back to dashboard
4.0/ 10 priority

Sources & remediation

Other references
Weakness type (CWE)
CWE-125

Priority

MEDIUM

Exploitation

ACTIVE

PoC

NONE

Patch

AVAILABLE

Momentum

STABLE

Threat summary

  • Active exploitation appears in 1 classified signals
  • Patch or workaround signal is available
  • 8 mentions across 6 observed days
  • Momentum state: stable

What's happening

  • Active exploitation reported across 1 signal
  • Patch or workaround mentioned in 2 signals
  • Technical details provided in 4 signals
  • General: 3 classified signals
  • Disclosure: 2 classified signals
  • Peaked 5d ago at 2 mentions (2026-05-19); latest day: 1
  • 8 total mentions across 6 days

Deep dive

Activity timeline8 mentions / 6d
01122Mentions · 2026-05-19: 2Mentions · 2026-09-18: 2Mentions · 2026-09-19: 1Mentions · 2026-09-20: 1Mentions · 2026-09-21: 1Mentions · 2026-10-03: 1Active Exploitation · 2026-09-21: 1Patch / Workaround · 2026-09-18: 1Patch / Workaround · 2026-09-21: 1Technical Details · 2026-05-19: 2Technical Details · 2026-09-18: 1Technical Details · 2026-09-21: 105-1909-1809-1909-2009-2110-03
Signal classification4 categories
General
342.9%
Disclosure
228.6%
Patch
114.3%
Active Exploitation
114.3%
Referenced assets10 URLs
Classification over time
DateTotalLabels
2026-05-192
Disclosure1General1
2026-09-182
Disclosure1Patch1
2026-09-191
General1
2026-09-201
General1
2026-09-211
Active Exploitation1
Full discourse8 posts
  • msuiche@msuiche
    General

    ELEGANTBOUNCER update to scan heic,heif,avif,hif files for CVE-2026-32741, CVE-2026-32882, CVE-2026-84383 + the no CVE exploit attempts in your own system. Great job again by @S1r1u5_ @rootxharsh & the rest of @HacktronAI team. https://t.co/ePLPDjnxp3

    Post summary

    The tweet announces an update to the ELEGANTBOUNCER scanner to detect exploitation attempts for three CVEs in specific image formats, crediting the HacktronAI team, without providing technical vulnerability details, PoCs, exploits, or patch information.

    42036305.6K
    2.2K followersView on X
  • مركز الأمن السيبراني للأبحاث والدراسات@ccforrs

    ثغـ...ـرة CVE-2026-32882 واستغلالها عبر Claude Opus 5: درس في هوية الدخول وأمن الصور التفاصيل .. https://url-shortener.me/PWAM #مركز_الأمن_السيبراني_للابحاث_والدراسات https://t.co/gU2KIWiGL2

    0101042
    44 followersView on X
  • hackerlogs@hackerlogs
    Active Exploitation

    The forum token reached Codex. Codex opened OpenAI's internal PR. @HacktronAI (13 Sep write-up; HN 19 Sep): chained a Discourse HEIC upload (CVE-2026-32882 / stale libheif) to an OpenAI SSO flaw on http://community.openai.com. Staff ChatGPT and Codex sessions followed. Proof: one employee's GitHub-connected Codex opened PR 1186742 in openai/openai. No source read, no merge, no customer data per the team. @OpenAI fixed identity in ~14 hours and paid $6,500 for the SSO finding, not Discourse testing. @discourse: rebuild the image; web UI update is not enough. A community login that can spend an internal coding agent is the blast radius. Primary: https://www.hacktron.ai/blog/hacking-openai Write-up: https://hackerlogs.com/blog/hacktron-openai-forum-sso

    Post summary

    A security researcher demonstrated chaining a Discourse HEIC upload (CVE‑2026‑32882) with an OpenAI SSO flaw to gain access to internal Codex, resulting in an unauthorized PR and prompting OpenAI to quickly patch the SSO issue and pay a bounty; the incident confirms active exploitation.

    10000100
    15 followersView on X
  • Aviatrix Threat Research Center@aviatrixtrc
    General

    Researchers exploited CVE-2026-32882 in OpenAI's forum to compromise staff ChatGPT accounts via shared SSO. The attack demonstrates how public services become pivot points for internal system access through compromised authentication. Runtime segmentation could have limited lateral movement between forum and staff services. 🔗 Full TRC analysis: https://aviatrix.ai/threat-research-center/claude-opus-5-openai-staff-account-takeover-cve-2026-32882

    Post summary

    Researchers demonstrated exploitation of CVE-2026-32882 in OpenAI's forum to compromise staff ChatGPT accounts via shared SSO, illustrating how public services can become pivot points for internal access, with a link to a full threat research analysis.

    0000052
    2.0K followersView on X
  • ♫Why♥Not♪@Python_s_
    Patch

    #تحذير ⚠️ 🚨 تحذير أمني من Discourse ثغرة CVE-2026-32882 في libheif قد تسمح بتنفيذ كود عن بُعد RCE عبر رفع صورة HEIF خبيثة إلى خوادم Discourse دون حاجة لتفاعل الضحية 🔴 الخطر: السيطرة على الخادم بصلاحيات التطبيق ولكنها حتى الآن ليست Zero-day مستغلة فعليًا ما لم يظهر دليل مستقل على ذلك ✅ الحل: التحديث فورًا إلى إصدار مصحح وإعادة بناء حاوية Discourse عبر ./launcher rebuild app المصدر الرسمي: Discourse Security Advisory https://github.com/discourse/discourse/security/advisories/GHSA-vhm9-85gw-x335?utm_source=chatgpt.com

    Post summary

    The post warns about CVE-2026-32882 in libheif affecting Discourse, noting it could allow RCE via malicious HEIF upload but is not actively exploited. It advises immediate update to a patched version and rebuilding the Discourse container.

    0000095
    222 followersView on X
  • Windows Forum@windowsforum
    Disclosure

    🔐 Not “Claude hacked OpenAI”: a public Discourse forum, HEIF flaw, and trusting SSO chained into employee ChatGPT, Codex, and source-code access. Security loves a three-step combo. https://windowsforum.com/news/cve-2026-32882-discourse-heif-flaw-led-to-openai-sso-access.444970/?utm_source=x&utm_medium=social&utm_campaign=news_node4 #Openai #SsoSecurity #Discourse #HeifVulnerability https://t.co/eNeA4Zqtf5

    Post summary

    Tweet announces CVE-2026-32882 describing a chain of Discourse forum exposure, HEIF flaw, and SSO trust issues leading to unauthorized OpenAI employee account/source-code access, linking to a full news article.

    0000061
    1.4K followersView on X
  • CVE@CVEnew
    Disclosure

    CVE-2026-32882 libheif is a HEIF and AVIF file format decoder and encoder. Versions 1.21.2 and prior contain a heap buffer over-read in HeifPixelImage::overlay() in libheif/pixelima… https://www.cve.org/CVERecord?id=CVE-2026-32882

    Post summary

    The text announces CVE-2026-32882 as a heap buffer over-read in libheif versions 1.21.2 and earlier, linking to the CVE record but providing no exploit or fix details.

    0000097
    57.5K followersView on X
  • Vulmon Vulnerability Feed@VulmonFeeds
    General

    CVE-2026-32882 Heap Buffer Over-Read in libheif HeifPixelImage Overlay Compositing Versions 1.21.2 https://vulmon.com/vulnerabilitydetails?qid=CVE-2026-32882

    Post summary

    A brief announcement of CVE-2026-32882 as a heap buffer over‑read issue in libheif, with only minimal technical detail and a reference link.

    0000049
    4.0K followersView on X

Explore more

Back to dashboard