CWE VIEW: Weaknesses Related to AI/ML Products
 Objective Audience
RelationshipsThe following graph shows the tree-like relationships between
weaknesses that exist at different levels of abstraction. At the highest level, categories
and pillars exist to group weaknesses. Categories (which are not technically weaknesses) are
special CWE entries used to group weaknesses that share a common characteristic. Pillars are
weaknesses that are described in the most abstract fashion. Below these top-level entries
are weaknesses are varying levels of abstraction. Classes are still very abstract, typically
independent of any specific language or technology. Base level weaknesses are used to
present a more specific type of weakness. A variant is a weakness that is described at a
very low level of detail, typically limited to a specific language or technology. A chain is
a set of weaknesses that must be reachable consecutively in order to produce an exploitable
vulnerability. While a composite is a set of weaknesses that must all be present
simultaneously in order to produce an exploitable vulnerability.
Show Details:
1448 - Weaknesses Related to AI/ML Products
Category - a CWE entry that contains a set of other entries that share a common characteristic.
Weaknesses That are Specific to AI/ML Technology
- (1446)
This category identifies weaknesses that are uniquely applicable to AI/ML technology.
Class - a weakness that is described in a very abstract fashion, typically independent of any specific language or technology. More specific than a Pillar Weakness, but more general than a Base Weakness. Class level weaknesses typically describe issues in terms of 1 or 2 of the following dimensions: behavior, property, and resource.
Inadequate Detection or Handling of Adversarial Input Perturbations in Automated Recognition Mechanism
- (1039)
The product uses an automated mechanism such as machine learning to recognize complex data inputs (e.g. image or audio) as a particular concept or category, but it does not properly detect or handle inputs that have been modified or constructed in a way that causes the mechanism to detect a different, incorrect concept.
Base - a weakness that is still mostly independent of a resource or technology, but with sufficient details to provide specific methods for detection and prevention. Base level weaknesses typically describe issues in terms of 2 or 3 of the following dimensions: behavior, property, technology, language, and resource.
Improper Validation of Generative AI Output
- (1426)
The product invokes a generative AI/ML
component whose behaviors and outputs cannot be directly
controlled, but the product does not validate or
insufficiently validates the outputs to ensure that they
align with the intended security, content, or privacy
policy.
Base - a weakness that is still mostly independent of a resource or technology, but with sufficient details to provide specific methods for detection and prevention. Base level weaknesses typically describe issues in terms of 2 or 3 of the following dimensions: behavior, property, technology, language, and resource.
Improper Neutralization of Input Used for LLM Prompting
- (1427)
The product uses externally-provided data to build prompts provided to
large language models (LLMs), but the way these prompts are constructed
causes the LLM to fail to distinguish between user-supplied inputs and
developer provided system directives.
prompt injection
Base - a weakness that is still mostly independent of a resource or technology, but with sufficient details to provide specific methods for detection and prevention. Base level weaknesses typically describe issues in terms of 2 or 3 of the following dimensions: behavior, property, technology, language, and resource.
Insecure Setting of Generative AI/ML Model Inference Parameters
- (1434)
The product has a component that relies on a
generative AI/ML model configured with inference parameters that
produce an unacceptably high rate of erroneous or unexpected
outputs.
Category - a CWE entry that contains a set of other entries that share a common characteristic.
General Software Weaknesses that Appear in Products that Use or Support AI/ML Technology
- (1447)
This category lists general software weaknesses in software that insecurely uses AI/ML components, but frequently appear in many kinds of software products that do not use AI/ML.
Base - a weakness that is still mostly independent of a resource or technology, but with sufficient details to provide specific methods for detection and prevention. Base level weaknesses typically describe issues in terms of 2 or 3 of the following dimensions: behavior, property, technology, language, and resource.
Improper Limitation of a Pathname to a Restricted Directory ('Path Traversal')
- (22)
The product uses external input to construct a pathname that is intended to identify a file or directory that is located underneath a restricted parent directory, but the product does not properly neutralize special elements within the pathname that can cause the pathname to resolve to a location that is outside of the restricted directory.
Path traversal
Directory traversal
Path transversal
Class - a weakness that is described in a very abstract fashion, typically independent of any specific language or technology. More specific than a Pillar Weakness, but more general than a Base Weakness. Class level weaknesses typically describe issues in terms of 1 or 2 of the following dimensions: behavior, property, and resource.
Improper Neutralization of Special Elements used in a Command ('Command Injection')
- (77)
The product constructs all or part of a command using externally-influenced input from an upstream component, but it does not neutralize or incorrectly neutralizes special elements that could modify the intended command when it is sent to a downstream component.
Command injection
Base - a weakness that is still mostly independent of a resource or technology, but with sufficient details to provide specific methods for detection and prevention. Base level weaknesses typically describe issues in terms of 2 or 3 of the following dimensions: behavior, property, technology, language, and resource.
Improper Neutralization of Special Elements used in an OS Command ('OS Command Injection')
- (78)
The product constructs all or part of an OS command using externally-influenced input from an upstream component, but it does not neutralize or incorrectly neutralizes special elements that could modify the intended OS command when it is sent to a downstream component.
Shell injection
Shell metacharacters
OS Command Injection
Base - a weakness that is still mostly independent of a resource or technology, but with sufficient details to provide specific methods for detection and prevention. Base level weaknesses typically describe issues in terms of 2 or 3 of the following dimensions: behavior, property, technology, language, and resource.
Improper Neutralization of Input During Web Page Generation ('Cross-site Scripting')
- (79)
The product does not neutralize or incorrectly neutralizes user-controllable input before it is placed in output that is used as a web page that is served to other users.
XSS
HTML Injection
Reflected XSS / Non-Persistent XSS / Type 1 XSS
Stored XSS / Persistent XSS / Type 2 XSS
DOM-Based XSS / Type 0 XSS
CSS
Base - a weakness that is still mostly independent of a resource or technology, but with sufficient details to provide specific methods for detection and prevention. Base level weaknesses typically describe issues in terms of 2 or 3 of the following dimensions: behavior, property, technology, language, and resource.
Improper Control of Generation of Code ('Code Injection')
- (94)
The product constructs all or part of a code segment using externally-influenced input from an upstream component, but it does not neutralize or incorrectly neutralizes special elements that could modify the syntax or behavior of the intended code segment.
Code Injection
Variant - a weakness that is linked to a certain type of product, typically involving a specific language or technology. More specific than a Base weakness. Variant level weaknesses typically describe issues in terms of 3 to 5 of the following dimensions: behavior, property, technology, language, and resource.
Improper Neutralization of Directives in Dynamically Evaluated Code ('Eval Injection')
- (95)
The product receives input from an upstream component, but it does not neutralize or incorrectly neutralizes code syntax before using the input in a dynamic evaluation call (e.g. "eval").
Class - a weakness that is described in a very abstract fashion, typically independent of any specific language or technology. More specific than a Pillar Weakness, but more general than a Base Weakness. Class level weaknesses typically describe issues in terms of 1 or 2 of the following dimensions: behavior, property, and resource.
Improper Encoding or Escaping of Output
- (116)
The product prepares a structured message for communication with another component, but encoding or escaping of the data is either missing or done incorrectly. As a result, the intended structure of the message is not preserved.
Output Sanitization
Output Validation
Output Encoding
Base - a weakness that is still mostly independent of a resource or technology, but with sufficient details to provide specific methods for detection and prevention. Base level weaknesses typically describe issues in terms of 2 or 3 of the following dimensions: behavior, property, technology, language, and resource.
Execution with Unnecessary Privileges
- (250)
The product performs an operation at a privilege level that is higher than the minimum level required, which creates new weaknesses or amplifies the consequences of other weaknesses.
Excessive Agency
Base - a weakness that is still mostly independent of a resource or technology, but with sufficient details to provide specific methods for detection and prevention. Base level weaknesses typically describe issues in terms of 2 or 3 of the following dimensions: behavior, property, technology, language, and resource.
Unrestricted Upload of File with Dangerous Type
- (434)
The product allows the upload or transfer of dangerous file types that are automatically processed within its environment.
Unrestricted File Upload
Base - a weakness that is still mostly independent of a resource or technology, but with sufficient details to provide specific methods for detection and prevention. Base level weaknesses typically describe issues in terms of 2 or 3 of the following dimensions: behavior, property, technology, language, and resource.
Deserialization of Untrusted Data
- (502)
The product deserializes untrusted data without sufficiently ensuring that the resulting data will be valid.
Marshaling/Marshalling, Unmarshaling/Unmarshalling
Pickling, Unpickling
PHP Object Injection
Class - a weakness that is described in a very abstract fashion, typically independent of any specific language or technology. More specific than a Pillar Weakness, but more general than a Base Weakness. Class level weaknesses typically describe issues in terms of 1 or 2 of the following dimensions: behavior, property, and resource.
Missing Authorization
- (862)
The product does not perform an authorization check when an actor attempts to access a resource or perform an action.
AuthZ
Base - a weakness that is still mostly independent of a resource or technology, but with sufficient details to provide specific methods for detection and prevention. Base level weaknesses typically describe issues in terms of 2 or 3 of the following dimensions: behavior, property, technology, language, and resource.
Server-Side Request Forgery (SSRF)
- (918)
The web server receives a URL or similar request from an upstream component and retrieves the contents of this URL, but it does not sufficiently ensure that the request is being sent to the expected destination.
XSPA
SSRF
Base - a weakness that is still mostly independent of a resource or technology, but with sufficient details to provide specific methods for detection and prevention. Base level weaknesses typically describe issues in terms of 2 or 3 of the following dimensions: behavior, property, technology, language, and resource.
Improper Neutralization of Special Elements Used in a Template Engine
- (1336)
The product uses a template engine to insert or process externally-influenced input, but it does not neutralize or incorrectly neutralizes special elements or syntax that can be interpreted as template expressions or other code directives when processed by the engine.
Server-Side Template Injection / SSTI
Client-Side Template Injection / CSTI
Base - a weakness that is still mostly independent of a resource or technology, but with sufficient details to provide specific methods for detection and prevention. Base level weaknesses typically describe issues in terms of 2 or 3 of the following dimensions: behavior, property, technology, language, and resource.
Improper Validation of Generative AI Output
- (1426)
The product invokes a generative AI/ML
component whose behaviors and outputs cannot be directly
controlled, but the product does not validate or
insufficiently validates the outputs to ensure that they
align with the intended security, content, or privacy
policy.
Base - a weakness that is still mostly independent of a resource or technology, but with sufficient details to provide specific methods for detection and prevention. Base level weaknesses typically describe issues in terms of 2 or 3 of the following dimensions: behavior, property, technology, language, and resource.
Improper Neutralization of Input Used for LLM Prompting
- (1427)
The product uses externally-provided data to build prompts provided to
large language models (LLMs), but the way these prompts are constructed
causes the LLM to fail to distinguish between user-supplied inputs and
developer provided system directives.
prompt injection
Base - a weakness that is still mostly independent of a resource or technology, but with sufficient details to provide specific methods for detection and prevention. Base level weaknesses typically describe issues in terms of 2 or 3 of the following dimensions: behavior, property, technology, language, and resource.
Insecure Setting of Generative AI/ML Model Inference Parameters
- (1434)
The product has a component that relies on a
generative AI/ML model configured with inference parameters that
produce an unacceptably high rate of erroneous or unexpected
outputs.
Vulnerability Mapping Notes Notes References View Metrics
Content
History
|
||||||||||||
