0day vulnerabilities have become rubbish in the AI era.
🌐 Official Website: https://0day-rubbish.com/blog
Traditional vulnerability disclosure is broken. It's slow, bureaucratic, and ineffective. In the AI era, we can mass-produce 0days at scale—making individual vulnerabilities less valuable but more impactful when disclosed directly.
We believe event-driven security hardening is the most effective approach: only when vendors face real, exploitable threats do they prioritize fixes.
Our automated AI systems continuously scan for vulnerabilities across real-world software, identifying potential 0-days through pattern analysis, fuzzing, and intelligent code review.
Each finding undergoes manual validation. We develop working proof-of-concept exploits to confirm exploitability and assess real-world impact.
Every week — Monday or Tuesday — we disclose a new batch of verified, exploitable 0-day vulnerabilities we've discovered and validated:
- Full technical analysis and root cause
- Working PoC exploit code
- Affected versions and systems
- Impact assessment
- Recommended mitigations
No delays. No bureaucracy. Just facts.
To all vendors: We hope you can complete fixes before hackers exploit these vulnerabilities.
- Real-world impact only: We disclose only vulnerabilities that affect real-world systems with actual user bases
- No worthless targets: Non-exploitable vulnerabilities or devices with negligible user adoption are excluded—they're rubbish with zero value
- Speed over protocol: Direct disclosure drives faster action than traditional channels
- Proof over claims: Every disclosure includes working exploits
- Impact over quantity: Focus on high-severity, widely-deployed vulnerabilities
- Transparency: Full technical details, no hidden agendas
- Non-profit: Driven by passion for security research, not financial gain
We partner with:
- Top AI model providers advancing automated security research
- Security researchers exploring AI-powered discovery
Our automated vulnerability discovery leverages cutting-edge large language models from leading AI providers:
- Anthropic (Claude) - Deep security pattern recognition and reasoning
- OpenAI - Advanced reasoning and code analysis
- DeepSeek - Specialized vulnerability detection
- Z.ai (GLM) - Long-context code analysis
- Moonshot (Kimi) - Long-context security analysis
An AI-driven research process (multi-LLM ensemble: Claude, OpenAI, DeepSeek, GLM, Kimi) discovers 0-days in real-world enterprise software. Every advisory below ships a full root-cause analysis plus a working, reproducible exploit script — no detection-only writeups, no withheld details.
Command execution reached from inside the product's own management surface — an unauthenticated report-upload page, a data-pipeline stage that compiles Scala, an energy gateway's event engine, a package-manager property, a media CGI, a banner uploader and a log viewer.
| # | Product | Affected Version | CVSS | Class | Advisory & PoC |
|---|---|---|---|---|---|
| 1 | Advantech WebAccess Node (SCADA/HMI) | 9.2.3 | 9.8 * | Unauthenticated CrystalRpt.aspx file upload with path traversal → code execution in the IIS worker process |
CrystalRpt.aspx → webshell → code execution |
| 2 | StreamSets Transformer | 3.17.0 | 8.1 * | ScalaDTransform code field compiled and run by the full Scala compiler in the service JVM, no sandbox → container root |
Scala stage → compiler → root |
| 3 | Circutor LineEds (industrial energy gateway) | 24.11.14-r0 | 9.8 * | Unauthenticated events.xml write with a shellExecute action plus a restart trigger → system() in pwrstudio |
events.xml shellExecute → system() |
| 4 | IPConfigure Orchid VMS | 26.3.0 | 7.2 * | Authenticated gpg_key server property → DNF rpm --import {} → uid 0 |
server property → rpm --import → root |
| 5 | Asustor ADM (AS602T) | 3.5.9.RWM1 | 8.8 * | Authenticated music.cgi act=live → As_System = system() |
music.cgi act=live → As_System |
| 6 | Maian Cart | 3.8 | 7.2 * | Authenticated addBanners upload with no extension allow-list → webshell in a served directory |
banner upload → PHP webshell |
| 7 | RCDevs WebADM (Freeware) | 2.4.14 | 7.2 | Authenticated logfile_viewer.php sid → popen, executing as webadm (uid 999) on an IAM/MFA platform |
sid → popen → webadm |
* More than one reading is published. Each of these advisories carries every reading alongside its own full vector, and labels what was verified versus what is conditional, inferred or rejected:
- Finding 1 — primary 9.8 at
PR:N, resting on the artifact-proven absence of any application-level authorization in the page plus the structural argument that the product's own login page must be anonymous. The advisory states plainly that no extracted installer artifact proves IIS anonymous authentication is enabled for that application path specifically, and publishes 8.8 atPR:Las the reading where it is not.- Finding 2 — primary 8.1 at
AC:H, because the shipped configuration file setshttp.authentication=aster; anonymous access requires thenonemode, which is a documented operator-selectable value and a code-level fallback rather than the shipped default. 9.8 atAC:Lis published for deployments where that mode is in effect.- Finding 3 — primary 9.8 at
PR:Ngrounded on a census of the authorization gate's 197 call sites, neither of which is on the vulnerable path. The record's "security is disabled by default" assertion is published as unverified rather than as an artifact-proven factory setting, and 8.8 atPR:Lis carried for the case where the gate applies.- Finding 4 — primary 7.2 at
PR:H: the shipped OpenAPI specification grants theconfigscope required by this endpoint to the Administrator role only. 9.8 is the reading where the installer's default administrator password is unchanged, which the packaged debconf template does evidence. The research record's 8.8 atPR:Lis disclosed and explained as not adopted.- Finding 5 — primary 8.8 at
PR:L, the only session validity check standing between the caller and the handler, with no role or privilege attribute consulted on that path. 7.2 atPR:His published as a conditional, untested reading.- Finding 6 — primary 7.2 at
PR:H, administrator session required. 9.8 atPR:Nis labelled as this advisory's own construction, conditional on deployment credentials that were never shown to be guessable.- Finding 7 — 7.2 is the only score published. The research record carried 8.8 against the same
PR:Hvector, which does not compute to 8.8; that is disclosed, andPR:L/8.8 is left as an open question for the vendor rather than adopted.
Totals: 7 advisories · 7 vendors · 3 in the unauthenticated class · 4 authenticated deep chains · every finding reaches operating-system command execution inside the product's own service or handler process. Executing identity is reported as measured where it was measured — uid 0 in a container (findings 2 and 4), webadm uid 999 (finding 7), uid 0 in a loopback lab process (finding 6) — and as a labelled inference where the device-side or shipped-pool identity was not measured (findings 1, 3, 5). No advisory prints a score its own vector does not compute to.
How each finding was verified — stated here because the seven differ, and because an advisory that blurs this is worth less than one that doesn't:
- 4 driven over HTTP against a real running instance of the product — the official container image (finding 2), a real
dpkg -iinstallation of the vendor package inside a Linux container (finding 4), a genuine installation served by the PHP built-in server (finding 6), and the vendor's Docker image over TLS (finding 7). Each of these reached the service from the same host over a loopback-bound port; none was driven from a separate network host. - 1 against shipped components deployed into a real IIS host over loopback (finding 1): the vulnerable code ran unmodified, but a stock installation was not achieved, so the public URL prefix is derived from a literal inside the decompiled page handler rather than from a request against a stock install.
- 2 with no live product traffic at all — the shipped ARM binary under
qemu-arm-staticwith the sink invoked through a debugger, and no HTTP request ever delivered (finding 3); and static analysis plus a faithful C replication of the sink run on an analysis host, with no physical device (finding 5).
Each advisory says which applies in its own section 9 and repeats it in its summary, and none presents an emulated, replicated or component-level run as an exploit against a live deployed device.
Earlier batches: Batch #1 · Batch #2 · Batch #3 · Batch #4 · Batch #5 · Batch #6 · Batch #7 · Batch #8 · Batch #9 · Batch #10 · Batch #11 · Batch #12
This is a continuous disclosure series. Thanks to continuous optimization, the AI-driven discovery pipeline now produces new 0-day findings at a stable daily rate, and we disclose verified batches on a weekly cadence.
- Latest batch: Batch 13 — 7 advisories; cumulative 121 across 13 batches
- Next drop: every Monday or Tuesday
- Future scope: expanding beyond enterprise IT into ICS / SCADA, energy, and aerospace systems
If you want to catch the next drop the moment it lands:
⭐ Star to bookmark · 👁 Watch (custom → Releases + Discussions) for new batches · 🌐 Follow the blog for per-advisory updates.
All disclosed vulnerabilities follow a standardized directory structure:
product/
└── <vendor>/
└── <version>/
└── <vulnerability_type>/
├── exploit/ # Exploit scripts and PoC code
├── analysis.md # Detailed vulnerability analysis
└── summary.md # Brief vulnerability overview
- product/: Root directory for all vulnerabilities
- /: Vendor or product name (e.g.,
apache,cisco,sonicwall) - /: Affected version range (e.g.,
6.11.0,12.4.2) - <vulnerability_type>/: Classification (e.g.,
unauth-rce,auth-bypass,deserialization-rce)
- exploit/: Directory containing working exploit scripts and PoC code
- analysis.md: Comprehensive technical analysis including root cause, attack vector, and impact
- summary.md: Concise vulnerability overview with affected versions and quick mitigation steps
product/
└── sonicwall/
└── sma-12.4/
└── preauth-deserialization-rce/
├── exploit/
│ └── poc.py
├── analysis.md
└── summary.md
Join us in redefining vulnerability disclosure for the AI era.