CVE-2024-5184
Official description Straight from the sourceThe vendor's or NVD's own wording, published unedited. Authoritative, but often terse — it says what broke, rarely what to do.
NVD · uneditedThe EmailGPT service contains a prompt injection vulnerability. The service uses an API service that allows a malicious user to inject a direct prompt and take over the service logic. Attackers can exploit the issue by forcing the AI service to leak the standard hard-coded system prompts and/or execute unwanted prompts. When engaging with EmailGPT by submitting a malicious prompt that requests harmful information, the system will respond by providing the requested data. This vulnerability can be exploited by any individual with access to the service.
Technical summary Written by usOur analysis, written from the advisory, the CVSS vector and the affected-version data. It adds context the advisory leaves out, and never invents facts that are not in the source.
dbcve analysis · high confidenceEmailGPT service has a direct prompt injection vulnerability where malicious users can submit crafted prompts to hijack the AI service logic. Attackers can force the system to leak hard-coded system prompts and execute arbitrary prompts, effectively taking control of the service behavior.
Verify against the referenced sources before acting — the references below are authoritative for this CVE, this summary is not.
Affected products & versions What the vendor confirmedThe version ranges the vendor confirmed as vulnerable. If your version sits inside a range here, treat yourself as exposed until you have upgraded.
NVD · CPE dataall versionsCVSS breakdown How the score is builtThe industry scoring standard. It rates how the flaw is reached, what it takes to exploit, and what an attacker gains — the score is derived from those, not the other way round.
From the vector- Attack vector
- Network
- Complexity
- Low
- Privileges
- None
- User interaction
- None
- Scope
- Unchanged
- Confidentiality
- High
- Integrity
- None
- Availability
- High
CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:N/A:H
Am I affected? How to checkSteps we derive from the advisory and the affected-version data, so you can decide whether this CVE reaches your setup. They are a guide, not a scan — your own configuration is the authority.
dbcve checksWork through these to decide whether this CVE applies to you.
-
Confirm Emailgpt is deployedLocate Emailgpt service installation by checking running processes, installed packages, or service configurations that reference 'Emailgpt'Affected if Emailgpt service is present in the environment
-
Identify prompt handling mechanismExamine how user-submitted prompts are processed - locate the code or configuration that accepts and forwards prompts to the AI modelAffected if User input is directly concatenated or passed to the AI without validation
-
Check for input sanitizationInspect the code or configuration for any input validation, sanitization, or filtering logic applied to user prompts before they reach the AIAffected if No input validation or sanitization is implemented for user prompts
-
Verify prompt isolationReview how system prompts are constructed - look for whether user input is separated from system instructions via templating, boundaries, or isolation mechanismsAffected if System prompts and user input are combined without clear boundaries or isolation
-
Test for injection susceptibilitySubmit a test prompt containing known injection patterns (e.g., instructions to ignore previous instructions, or commands that could override system behavior) and observe if the AI responds to the injected contentAffected if The system executes or leaks content from injected prompt patterns
The environment is affected if Emailgpt is deployed and user prompts are passed directly to the AI model without sanitization, input validation, or prompt isolation from system instructions.
Generated from the published advisory. Verify against your own configuration.
Remediation Closing itWhat it takes to close this. Where a vendor fix exists we point at it; where none exists we say so plainly, and can build one. Effort estimates are scoped from the advisory, not from your codebase.
From vendor dataImplement strict input validation and sanitization for all user prompts. Redesign prompt handling to use prompt isolation techniques such as separating user input from system prompts, using prompt templating with clear boundaries, or implementing guardrails that detect and block injection attempts.
- Consultation6.0 h
- Implementation16.0 h
- Testing12.0 h
- Review / QA6.0 h
An estimate, not a bill — we confirm scope with you before any work starts. Need it this week? Rush from $11,136.
Scan for this in your stack
Free · runs locallyCheck whether your project pulls in CVE-2024-5184 — or any other known-vulnerable package — straight from your lock files. Free and open source; it runs locally and uploads nothing.
References Go to the primary sourcePrimary sources — vendor advisories, patches and trackers. Where our summary and a reference disagree, the reference wins.
Primary sourcesPractitioner notes
ContributedPeer-ranked notes from engineers who’ve handled CVE-2024-5184 in production — separate from our analysis above.
The advisory tells you what broke. It rarely tells you what actually worked. If you’ve dealt with this one, that detail is what the next engineer is searching for.
- The version that genuinely resolved it — not the one the vendor claimed
- A config change or rule that shut the vector down
- A gotcha in the upgrade path that cost you an afternoon
No notes yet
Be the first to add a field note for this CVE — a mitigation you’ve verified, a version caveat, or a link to a working fix. Sign in above to contribute.
A place for practitioners to share what actually worked: a mitigation you’ve tested, a configuration change, a version- or environment-specific caveat, or a link to a verified patch. The most useful notes rise to the top as peers upvote them, so the signal stays high.
- Verified mitigations, workarounds, and config changes
- Version or environment caveats, and links to real fixes
- No weaponised exploit code, or anything meant to cause harm
- No spam, self-promotion, credentials, or personal data