惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

GbyAI
GbyAI
Simon Willison's Weblog
Simon Willison's Weblog
Microsoft Security Blog
Microsoft Security Blog
Y
Y Combinator Blog
The GitHub Blog
The GitHub Blog
Engineering at Meta
Engineering at Meta
F
Fortinet All Blogs
OSCHINA 社区最新新闻
OSCHINA 社区最新新闻
A
About on SuperTechFans
Last Week in AI
Last Week in AI
月光博客
月光博客
有赞技术团队
有赞技术团队
P
Proofpoint News Feed
MyScale Blog
MyScale Blog
Martin Fowler
Martin Fowler
钛媒体:引领未来商业与生活新知
钛媒体:引领未来商业与生活新知
C
Check Point Blog
U
Unit 42
The Register - Security
The Register - Security
奇客Solidot–传递最新科技情报
奇客Solidot–传递最新科技情报
Cyber Security Advisories - MS-ISAC
Cyber Security Advisories - MS-ISAC
Hugging Face - Blog
Hugging Face - Blog
阮一峰的网络日志
阮一峰的网络日志
V
Visual Studio Blog
酷 壳 – CoolShell
酷 壳 – CoolShell
D
DataBreaches.Net
WordPress大学
WordPress大学
aimingoo的专栏
aimingoo的专栏
H
Hacker News: Front Page
Recent Announcements
Recent Announcements
C
CXSECURITY Database RSS Feed - CXSecurity.com
Latest news
Latest news
小众软件
小众软件
P
Palo Alto Networks Blog
PCI Perspectives
PCI Perspectives
Security Latest
Security Latest
S
Secure Thoughts
Scott Helme
Scott Helme
Threat Intelligence Blog | Flashpoint
Threat Intelligence Blog | Flashpoint
T
Threat Research - Cisco Blogs
P
Proofpoint News Feed
M
MIT News - Artificial intelligence
Application and Cybersecurity Blog
Application and Cybersecurity Blog
Google DeepMind News
Google DeepMind News
Recorded Future
Recorded Future
O
OpenAI News
S
Securelist
云风的 BLOG
云风的 BLOG
H
Help Net Security
T
Troy Hunt's Blog

CSO Online

New malware turns Linux systems into P2P attack networks Poisoned truth: The quiet security threat inside enterprise AI Train like you fight: Why cyber operations teams need no-notice drills Die besten DAST- & SAST-Tools CISA mulls new three-day remediation deadline for critical flaws CISA pushes critical infrastructure operators to prepare to work in isolation CISOs step up to the security workforce challenge 10 Anzeichen für einen schlechten CSO Anthropic Mythos spurs White House to weigh pre-release reviews for high-risk AI models Security agencies draw red lines around agentic AI deployments The fake IT worker problem CISOs can’t ignore How CISOs should utilize data security posture management to inform risk Was ist ein Botnet? Human-centric failures: Why BEC continues to work despite MFA Just 34% of cyber pros plan to stick with their current employer Managing OT risk at scale: Why OT cyber decisions are leadership decisions 4 ways to prepare your SOC for agentic AI ‘Trivial’ exploit can give attackers root access to Linux kernel Bank regulator sounds warning over cybersecurity threat posed by AI models Dismantle implicit trust in OT networks, CISA tells critical infrastructure operators Max-severity RCE flaw found in Google Gemini CLI Stopping the quiet drift toward excessive agency with re-permissioning ODNI to CISOs on threat assessments: You’re on your own 10 wichtige Security-Eigenschaften: So setzen Sie die Kraft Ihres IT-Sicherheitstechnik-Teams frei Researchers unearth industrial sabotage malware that predated Stuxnet by 5 years AWS leans on prior ingenuity to face future AI and quantum threats What it takes to win that CSO role Third Party Risk Management: So vermeiden Sie Compliance-Unheil Critical Cursor bug could turn routine Git into RCE Securing RAG pipelines in enterprise SaaS What CISOs need to get right as identity enters the agentic era Stopping AiTM attacks: The defenses that actually work after authentication succeeds EDR-Software – ein Kaufratgeber Microsoft patched an ‘agent-only’ role that was not AI is reshaping DevSecOps to bring security closer to the code The 'manager of agents': How AI evolves the SOC analyst role 4 Wege aus der Security-Akronymhölle Autonome KI-Agenten: Strategien für die neue Bedrohungslage New US House privacy bills raise hard questions about enterprise data collection Scattered Spider co-conspirator pleads guilty Security-KPIs und -KRIs: So messen Sie Cybersicherheit Bitwarden CLI password manager trojanized in supply chain attack 3 practical ways AI threat detection improves enterprise cyber resilience The curious case of Sean Plankey’s derailed CISA nomination Google gets agent-ready for the Mythos age Google drafts AI agents secure systems against AI hackers CNAPP – ein Kaufratgeber Riddled with flaws, serial-to-Ethernet converters endanger critical infrastructure NFC tap-to-pay gets tapped by hackers Anthropic bets on EPSS for the coming bug surge SBOM erklärt: Was ist eine Software Bill of Materials? Thousands of Apache ActiveMQ instances still unpatched, weeks after an actively exploited hole discovered Prompt injection turned Google’s Antigravity file search into RCE Why identity is the driving force behind digital transformation Top techniques attackers use to infiltrate your systems today The thin gray line: Handala, CyberAv3ngers and Iran’s proxy ops Attackers abuse Microsoft Teams to impersonate the IT helpdesk in a new enterprise intrusion playbook CISOs reshape their roles as business risk strategists Copilot & Agentforce offen für Prompt-Injection-Tricks Claude Mythos – ist der Hype gerechtfertigt? Für Cyberattacken gewappnet – Krisenkommunikation nach Plan Critical sandbox bypass fixed in popular Thymeleaf Java template engine White House moves to give federal agencies access to Anthropic’s Claude Mythos Another Microsoft Defender privilege escalation bug emerges days after patch Palo Alto’s Helmut Reisinger sees a cyber sea change ahead as AI advances Positiv denken für Sicherheitsentscheider: 6 Mindsets, die Sie sofort ablegen sollten NIST cuts down CVE analysis amid vulnerability overload Was bei der Cloud-Konfiguration schiefläuft – und wie es besser geht The endless CISO reporting line debate — and what it says about cybersecurity leadership Behind the Mythos hype, Glasswing has just one confirmed CVE Insurance carriers quietly back away from covering AI outputs RCE by design: MCP architectural choice haunts AI agent ecosystem Critical nginx UI tool vulnerability opens web servers to full compromise Copilot and Agentforce fall to form-based prompt injection tricks The deepfake dilemma: From financial fraud to reputational crisis 7 biggest healthcare security threats The need for a board-level definition of cyber resilience Mallory Launches AI-Native Threat Intelligence Platform, Turning Global Threat Data Into Prioritized Action 13 Fragen gegen Drittanbieterrisiken April Patch Tuesday roundup: Zero day vulnerabilities and critical bugs 4 questions to ask before outsourcing MDR 5 trends defining the future of AI-powered cybersecurity EU regulators largely denied access to Anthropic Mythos China-linked cloud credential heist runs on typos and SMTP How AI is transforming threat detection The AI inflection point: What security leaders must do now Cyber-Inspekteur: Hybride Attacken nehmen weiter zu Anthropic’s Mythos signals a structural cybersecurity shift Seven IBM WebSphere Liberty flaws can be chained into full takeover Old Docker authorization bypass pops up despite previous patch Hacker Unknown now known, named on Europol’s most-wanted list The cyber winners and losers in Trump’s 2027 budget CMMC compliance in the age of AI Claude uncovers a 13‑year‑old ActiveMQ RCE bug within minutes Was CISOs von Moschusochsen lernen können Hackers have been exploiting an unpatched Adobe Reader vulnerability for months New ClickFix variant bypasses Apple safeguards with one‑click script execution Cloudflare ‘actively adjusting’ quantum priorities in wake of Google warning Patch windows collapse as time-to-exploit accelerates So geht Post-Incident Review
Hole in widely-used FFmpeg codec could crash media servers or enable RCE
Howard Solomon · 2026-06-24 · via CSO Online

Research from JFrog into the software supply chain vulnerability points to the need for better visibility into applications, including SBOMs.

A newly discovered critical vulnerability in the FFmpeg media processing framework bundled in a huge number of open source and commercial applications points, again, to the need for CSOs to have strategies to deal with software supply chain vulnerabilities, which should include demanding a software bill of materials for all products.

Found by researchers at JFrog, the hole (CVE-2026-8461) is a heap out-of-bounds write in the MagicYUV decoder that can crash any application that uses the framework. It runs in everything from desktop video players like Kodi and mpv, to Linux file-manager thumbnail generators, to cloud transcoding pipelines (such as AWS MediaConvert and Cloudflare Stream) and self-hosted media servers.

The hole has been dubbed PixelSmash.

“The vulnerability can be used to crash systems and, in worst cases, can be escalated to remote code execution, meaning it should be taken seriously and prioritized by security teams and developers,”  Yuval Moravchik, JFrog’s vulnerability research team lead, said in an email. “CSOs and developers should make sure their application security products alert them to the presence of this vulnerability, the sooner the better.”

The researchers said they have demonstrated the full exploit, achieving remote code execution on two independent targets: a Jellyfin media server (via automatic library scan) and a Nextcloud collaboration platform instance (via the video preview provider), in both cases by simply uploading a crafted 50 KB AVI file.

In fact, any crafted media file (AVI, MKV, or MOV container) will work on an application that uses FFmpeg’s libavcodec. Even a file folder containing the app is vulnerable, because the file manager’s thumbnail generator will trigger the bug.

“All it takes is processing a single malicious media file,” the researchers say.

There is one workaround: If the MagicYUV decoder is not needed, it can be disabled by developers at build time.

However, Garrett Calpouzos, principal security researcher at Sonatype, doubts full exploitation will be common. “I would be surprised to see broad, reliable exploitation of this specific bug in modern, hardened environments,” he told CSO in an email. “The more realistic near-term risk is denial-of-service (DoS), especially for services that process untrusted media at scale.”

Regardless, users of FFmpeg should upgrade to the patched version (8.1.2) as soon as possible if they know it’s in their application or are so informed by vendors.

A foundational dependency

JFrog notes FFmpeg is bundled with or linked to by virtually every media-processing application on every platform. It confirmed crashes against Kodi, mpv, ffmpegthumbnailer (used by GNOME, KDE, XFCE), Jellyfin, Emby, Nextcloud, Immich, PhotoPrism, and OBS Studio, among others.

The vulnerability is a single bug in the codec decoder inside FFmpeg, but it’s a foundational dependency embedded in hundreds of downstream projects that cascades to every application that links libavcodec, an open source library that provides the core encoding and decoding capabilities for audio and video streams.

None of the affected projects introduced this bug, the researchers note. They inherited it silently through their dependency on FFmpeg. And most, they add, have no mechanism to detect or mitigate it independently.

This isn’t the first security issue in FFmpeg. As researchers at DepthFirst pointed out earlier this month, Google’s Big Sleep team disclosed 13 vulnerabilities, and Anthropic, using its Claude Mythos preview model, found a 16-year-old hole. In April, researchers at SentinelOne described a buffer overflow vulnerability, and last December researchers at ZeroPath reported finding seven memory vulnerabilities.

Combatting supply chain vulnerabilities

Software supply chain vulnerabilities due to weaknesses in third party libraries and open source components have long been known as a security risk. Arguably the most infamous is the 2020 compromise of the update mechanism for SolarWinds Orion IT infrastructure management platform in which a Russian threat group called APT20/Cozy Bear installed a backdoor into a legitimate update that went to 18,000 customers, although a much smaller group was exploited.

To combat supply chain vulnerabilities, experts say developers need to adopt strategies to scrutinize code before it is deployed. These include software composition analysis, which gives visibility into the software’s dependencies, static application security testing, container scans, and having or generating a software bill of materials (SBOM).

[Related content: What is an SBOM?]

SBOMs critical

SBOMs are easy to create if a developer is building their own app. They’re harder to get from downloaded or commercial applications.

Johannes Ullrich, dean of research at the SANS Institute, told CSO that a transparent declaration of dependencies via SBOMs is critical if organizations are to accurately understand the risks posed by software. In particular, commercial software vendors are often hesitant to declare components; the perceived value that commercial software attempts to project is often incompatible with the wrappers it implements around commonly used open-source components.

One of the problems with the PixelSmash vulnerability, he pointed out, is the use of FFmpeg in an application is often neither obvious nor declared. An SBOM would help CSOs or heads of development teams quickly learn if any of their applications are affected.

What will it take to encourage CSOs to make SBOMs part of their security strategies? “Compliance regulation,” Ullrich replied. “These changes are usually only made if compliance requires them. Some influence may be exerted by government customers requiring SBOMs, but again, that will only happen if compliance requires this as part of purchasing guidelines.”

Lesson: Attack surface management

Sonatype’s Calpouzos said one big lesson for enterprises from the PixelSmash discovery is attack surface management. MagicYUV is a niche lossless video format used more in high-end video editing workflows than in mainstream web video delivery, he pointed out, and FFmpeg is typically built with every decoder enabled, meaning most applications end up exposing code paths they may never actually need. Infosec teams need to ensure they only enable the formats and features that their organization actually use in applications.

“This is also exactly where SBOMs matter,” he added. “Most organizations do not have a complete understanding of where FFmpeg is embedded, whether it is bundled or statically linked, or which optional features are enabled. An SBOM helps security teams move from ‘Are we exposed?’ to ‘Where are we exposed, and how fast can we fix it?’ In the AI era, attackers and researchers alike can increasingly comb through mature open source projects for overlooked vulnerabilities in obscure features, so organizations need to know what they ship, minimize what they expose, keep default security controls enabled, and patch quickly.”

Recommendations for SBOMs

The US Cybersecurity and Infrastructure Security Agency (CISA) has circulated a suggested list of minimum elements a software bill of materials should include.

“An effective mechanism for sharing and using software data must be machine-processable and scalable,” the it notes. “The SBOM model achieves both by capturing software component data in a machine-processable format and supporting operations that analyze, share, and manage it. SBOM data can be mapped to other data sources such as security advisories or organization-level ‘approved/not approved’ software databases to improve other priority practices (e.g., secure software development, vulnerability management). SBOM will not resolve all software security and supply chain concerns, but it is a necessary step that enables and empowers risk-informed security decision making.”

Separately, last month the G7 cybersecurity working group, which includes the US, Germany, Canada, France, Italy, Japan, the United Kingdom, and the European Union, released joint guidance, Software Bill of Materials for AI – Minimum Elements, to help public and private sector stakeholders improve transparency in their artificial intelligence (AI) systems and supply chains.

However, JFrog’s Moravchik argues that while a software bill of materials is an essential first step, it’s only a starting point to more secure applications. “The teams that stay ahead pair it with continuous CVE mapping and exploitability analysis, scan at the binary level where these dependencies actually live, and disable codecs and features they don’t use,” he told CSO.

Infosec leaders also need to shift from reacting to proactively gating, he said. That means moving security enforcement upstream so risk is blocked at the door, through automated governance of every package, model, and agentic tool entering the pipeline paired with AI-powered threat detection, rather than remediating in the wild after a CVE drops.

SUBSCRIBE TO OUR NEWSLETTER

From our editors straight to your inbox

Get started by entering your email address below.