143 results about "Injection attacks" patented technology

The embodiment of the invention relates to the field of artificial intelligence, and provides a model context protocol injection attack protection method and device based on dynamic semantic analysis, computer equipment and a storage medium, and the method comprises the steps: obtaining request data corresponding to a call request initiated by a user through a model context protocol, carrying out the preprocessing of the request data, obtaining the preprocessed request data; performing semantic vectorization on request text and context historical information in the request data through a lightweight bidirectional encoder representation model to output an initial risk score; correcting the initial risk score according to context historical information carried in a model context protocol to obtain a corrected final risk score; and performing hierarchical defense decision according to the final risk score, and determining risk grading information corresponding to each piece of request data so as to execute a protection action corresponding to each piece of risk grading information. By adopting the method, the accuracy of identifying the protocol injection attack can be improved.

The invention discloses a multi-modal attack identification method fusing BMama and difference to guide trans-attention, which comprises the following steps: simulating a false data injection attack, a denial of service attack, an address resolution protocol spoofing attack and a domain name system spoofing attack, collecting physical layer sensor data and network layer flow data, and preprocessing multi-modal data; bMama is constructed to perform dynamic time modeling on multi-modal data, a graph neural network is combined to adversariate a variational auto-encoder, features of a power grid system topology and a communication topology structure are fused, and robustness of potential representation is enhanced through adversarial training; the method comprises the following steps of: guiding feature complementary fusion by using modal difference through a difference guide iteration cross-attention fusion mechanism, improving the capability of distinguishing complex attacks, finally carrying out attack detection and classification on fused modals, and executing end-to-end optimization according to a weighted combination of loss of each part. The method can effectively detect and classify the multi-modal attack in the smart power grid, and enhances the safety and reliability of a complex system.

The invention discloses a power network attack chain dynamic deduction and intelligent response process method, system and device based on deep reinforcement learning and a medium, and belongs to the technical field of network security and power system protection. Multi-modal data is aligned and normalized, an event view cache is constructed, and the generalization detection capability on process camouflage and memory injection attacks is improved through a federated learning collaborative detection mechanism; based on the event view cache and historical threat intelligence, generating a dynamic attack knowledge graph, constructing a deep reinforcement learning model taking the attack knowledge graph as an environment, calculating an attack influence index by using a Bayesian network, and generating a differentiated security response instruction; and realizing attack path backtracking and attack source positioning based on the attack knowledge graph. According to the method, multi-modal data fusion analysis and strategy adaptive updating are realized, and the attack chain identification accuracy and evidence chain construction integrity are remarkably improved.

In an embodiment, a method includes receiving a prompt provided to a large language model (LLM), generating a semantic fingerprint for the prompt based on semantic and syntactic features associated with the prompt, generating a vector representation incorporating the semantic fingerprint for the prompt, calculating semantic distances between the vector representation and multiple vector representations associated with multiple malicious intents, determining an intent associated with the prompt based on the semantic distances, and determining a defense action on the prompt based on the intent and multiple policies to mitigate a security risk associated with the prompt.

A malicious instructions detection model (“detector”) intercepts augmented prompts destined for a large language model (“LLM”). Each augmented prompt was augmented with data from potentially compromised data sources susceptible to indirect prompt injection attacks. The detector tokenizes / preprocesses sentences in the augmented prompts and is invoked on the tokenized / preprocessed sentences to obtain confidence scores that each sentence comprises malicious instructions. If one or more of the confidence scores is above a threshold, the detector blocks the augmented prompt and generates an alert indicating the blocking and the malicious instructions. Otherwise, the detector communicates the augmented prompt to its intended LLM.

Prompt injection attack detection in responses from large language models includes receiving, at a server from a user device, a user prompt segment to an LLM, generating a LLM prompt from the user prompt segment, sending the LLM prompt to the LLM, and receiving a response from the LLM. Prompt injection attack detection further includes comparing the response to a structured data schema for the response to validate the response, and sending, responsive to validating the response, the response to the user device.

The invention discloses a false data injection attack detection method for an electric power information physical system, relates to the technical field of attack detection, and solves the problem that in the prior art, only the damage of FDIA to a measurement data time mode is considered, and the spatial correlation between measurement data is affected even if the FDIA is not considered. According to the method, an FDIA detection model structure based on correlation difference is designed, and the attention action range is constrained by embedding position information and introducing a correction factor. Besides, in combination with fine granularity and multi-scale characteristics of a measurement data sequence, topological correlation is introduced, correlation differences are defined, a correlation difference-based antagonism discriminant criterion reasoning method is designed, and the distinguishability of normal and attack measurement data is amplified through adversarial training, so that a discriminant criterion is obtained. Complete embodiments show that compared with an existing detection model, the provided model has higher detection accuracy and robustness, and is excellent in performance when facing injection attacks with different amplitudes.

The invention discloses an SQL injection detection method and device, equipment and a storage medium, and relates to the technical field of safety protection, and the method comprises the steps: obtaining an initial SQL statement input by a target user through a target Web application, and carrying out the preprocessing of the initial SQL statement, so as to obtain a corresponding target SQL statement; generating a corresponding token sequence based on the target SQL statement, and constructing a target abstract syntax tree corresponding to the target SQL statement; wherein the token sequence is a sequence generated based on each SQL element in the target SQL statement; and extracting a target feature corresponding to the target SQL statement based on the token sequence and the target abstract syntax tree, and detecting whether the target SQL statement has an injection attack behavior based on the target feature by using a pre-trained target graph neural network model. According to the method, high-accuracy, high-adaptability and low-delay SQL injection real-time detection can be realized.

The invention discloses a three-layer cooperative attack detection and defense method and device oriented to a large language model context protocol. Aiming at the technical bottleneck problems of low detection accuracy and poor real-time performance caused by the fact that an existing large language model is easily subjected to semantic attack when calling an external tool, a three-level protection assembly line is constructed: firstly, an explicit threat is intercepted at a millisecond-level speed by adopting a regular static scanner capable of being subjected to hot updating; secondly, accurately identifying hidden confrontation prompts and complex semantic attacks by utilizing a multi-language text embedding model which is subjected to full-parameter fine adjustment on MCP injection attack statements; and finally, introducing an independent and lightweight LLM to carry out confidence fusion rechecking on a detection result. Experimental results show that the detection accuracy, the recall rate, the false alarm suppression, the response real-time performance, the cross-domain generalization ability to non-attack styles and the like are all remarkably improved, and continuous and reliable safety protection can be provided for a large language model tool chain.

The invention relates to the technical field of artificial intelligence and information security, provides a cue word injection attack detection method and device of a large language model, equipment, a storage medium and a program product, and has a better cue word injection attack detection effect. The method comprises the following steps: generating cue words from preset test corpora, and inputting the cue words into a first large language model; the test corpus generation cue word is used for indicating the first large language model to generate a cue word injection test corpus in combination with multiple cue word injection modes for each cue word injection attack type; the cue word injection test corpus is input into the to-be-tested large language model, and an input and output data pair is obtained according to response content output by the to-be-tested large language model; and calling the second large language model to detect whether the to-be-detected large language model is successfully attacked according to the input and output data to obtain a target detection result aiming at the cue word injection attack.

The invention relates to a scene self-adaptive white-box prompt injection attack system. The system comprises a self-adaptive context prompt generation module, an enhanced GCG algorithm-based confrontation suffix generation module and a simulator-based intelligent agent test module with a body. The adaptive context prompt generation module constructs specific prompt templates for different scenes, and automatically identifies and encodes unique semantic features of different application scenes. And the adversarial suffix generation module based on the enhanced GCG algorithm calculates the coordinate gradient of each token position, provides direction guidance for subsequent token replacement based on a double-component loss function, and adopts a weighted combination loss function to realize system optimization. The simulator-based intelligent agent test module carries out semantic understanding and intention recognition through a target large language model, converts a natural language into a control instruction, recognizes different types of commands, and maps the analyzed commands to a specific API for calling.

The invention relates to an elastic control method of a high-order nonlinear system under pulse false information injection attack. The method comprises the following steps: acquiring a system output signal and a control input signal of a current control period, and a state estimation value vector and a fuzzy weight vector estimation value of a previous control period; inputting a system output signal and a control input signal of a current control period, and a state estimation value vector and a fuzzy weight vector estimation value of a previous control period into a nonlinear state observer based on a fuzzy logic system to obtain a state estimation value vector of the current control period; the controller based on the fuzzy logic system carries out backstepping recursion through the state estimation value vector of the current control period and the fuzzy weight vector estimation value of the previous control period to obtain an elastic control input signal; and obtaining a fuzzy weight vector estimator of the current control period through an adaptive law based on the state estimation value vector of the current control period. By adopting the method, state jump and performance damage of attacks can be resisted.

The invention discloses an active defense security control method for a hidden false data injection attack, and relates to the field of information physical system security control, in particular to an active defense security control method. The invention aims to solve the problem that the existing detection method can encrypt a transmitted signal, but is limited to detection of a specific type of attack, or dynamic coupling changing the configuration of a control system can increase the burden of a state estimator, and even causes the performance reduction of the system. The method comprises the following steps of: 1, acquiring output feedback gains of the information physical system under different switching signals; and 2, based on the output feedback control gain obtained in the step 1, performing active attack detection and positioning on the FDI attack, and when the attack is detected and positioned, performing active switching by a controller of the information physical system so as to construct an active security controller aiming at the FDI attack.

The invention belongs to the technical field of computer information security, particularly discloses a content security protection method and device based on semantic consistency, and aims to solve the problem that high-level cue word injection attacks are difficult to effectively recognize in the prior art. Carrying out real-time interception on dominant illegal texts through a content filtering module; quantizing generation rationality differences of prompt words between attack and normal language models by using a word vector confusion degree calculation module; evaluating the logic coherence of the sentence structure of the prompt word through a statement semantic consistency judgment module; integrating the multi-dimensional feature data and carrying out risk classification by adopting a machine learning algorithm; and routing the suspected attack request to a value fine tuning model for processing according to a judgment result. According to the technical scheme, high-precision recognition and response to complex cue word injection attacks are achieved, and the content safety protection capacity and compliance guarantee level of a large model in an interaction scene are remarkably improved.

The invention relates to a node injection attack method based on adaptive target selection, and the method comprises the following steps: S1, target node selection: calculating a comprehensive score of a node based on uncertainty and topology centrality, and dynamically selecting a target node set of a current attack round; s2, feature generation: using an adaptive feature generator to generate node features which are similar to target node distribution and have strong aggressiveness; and S3, disturbance edge construction: selecting an optimal disturbance edge connection mode for the injection node according to strategy network output in reinforcement learning. According to the method, the attack flexibility can be improved through dynamic target selection, the attack performance can be remarkably enhanced through combination of disturbance characteristics and structures, and the method has good concealment, expandability and generalization ability. The method is widely applied to security evaluation and defense research fields related to graph neural networks, such as social network analysis, recommendation systems, knowledge graphs and the like.

The application discloses a large language model prompt word injection attack detection method and device based on context learning, and belongs to the technical field of artificial intelligence, and comprises the following steps: based on a Bert pre-training model, inherent features and dependency relationships between different levels and different labels are learned by introducing a label-based attention module, a multi-level, multi-label and fine-grained classification model for prompt word injection attack is designed, and accurate identification of the prompt word injection attack is realized.Meanwhile, according to the context learning, the prediction ability of the classification model is combined with the ability of the large language model, and the defense ability of the large language model to the prompt word injection attack is improved.The application can automatically detect the prompt word injection attack, improve the effectiveness and comprehensiveness of detection, and can be effectively applied to the field of large model security detection.

Computer-implemented methods and systems for training an adversarial neural network to simulate a stealthy blind false data injection attack on a cyber physical system are provided. Supervised learning is used to generate an initial attack vector, by an adversarial attack generation model, based on inferred grid topology and historical measurements. A final attack vector is generated, by an adversarial verification model, based on a filtered subset of the initial attack vector utilizing a substitute bad data detection threshold, wherein the final attack vector enables creation of a counter measure.

The application discloses a false data injection attack under unmanned cluster system attack detection device and method, belongs to information security technical field, the device includes: encryption module, decryption module, data similarity calculation module, threshold comparator and alarm; Encryption module is used for carrying out encryption processing to the measurement output signal, and the encrypted signal is transmitted to the decryption module; Decryption module is used for decrypting the encrypted signal; Data similarity calculation module is used for calculating the real-time residual signal of the system by using the decrypted signal, and calculating the probability distribution difference between the real-time residual signal of the system and the residual signal when the system is normally operated; Threshold comparator is used for comparing the calculated probability distribution difference with the preset threshold, if greater than the preset threshold, then trigger the alarm. The application can consume less energy, quickly detect the false data injection attack, so as to guarantee the system safety.

The invention discloses a vulnerability analysis method of a multi-time-scale micro-grid under FDI attack, which comprises the following steps: decomposing system dynamics of a direct current micro-grid into a fast boundary layer subsystem and a slow order reduction subsystem by using a singular perturbation theory, respectively deducing stability conditions of each subsystem under a false data injection attack condition, and analyzing the vulnerability of the multi-time-scale micro-grid under the false data injection attack condition; and the input state stability of the low-speed subsystem under the FDI attack is proved. The worst deviation of the system state under the attack condition is calculated by constructing a zootope reachable set analysis framework, and the maximum allowable attack amplitude capable of ensuring safe operation of the system is quantified. According to the method, through numerical simulation and hardware experiment verification, a novel attack surface introduced by multi-time scale characteristics can be effectively revealed, and theoretical support and guidance are provided for safe and stable control of the DC micro-grid.

This invention discloses a method and system for preventing false information injection using a large language model. The method involves: dynamically evaluating the credibility of user input to obtain an input credibility score; calculating a user reputation score based on historical user behavior data and mapping it to a generation permission level; retrieving authoritative knowledge fragments from a closed-loop trusted knowledge graph for the input and constructing generation constraint instructions based on the permission level; calling the large language model to generate response content according to the permission level and constraint instructions; performing consistency verification on the response content and then outputting it; finally, generating a lineage log containing end-to-end interaction data and storing it on the blockchain. This invention, by integrating input credibility assessment, user reputation coupling, knowledge tracing constraints, and end-to-end auditing, achieves pre-emptive identification, process blocking, and post-event traceability of false information, effectively solving the problem of the lack of systematic defense against false information injection attacks in existing technologies, and significantly improving the content security and compliance of generative artificial intelligence applications in key areas.

The invention relates to a low-overhead moving target defense method and device for false data injection attacks, and the method comprises the steps: constructing a topological structure of a power grid bus and branches based on actual intelligent power grid data, and associating the branches with an initial admittance value; establishing an initial measurement matrix based on the initial admittance value and the topological structure of the power grid bus and the branches, and constructing a moving target defense strategy based on the initial measurement matrix; based on an invalid branch identification criterion and a topological structure of a power grid bus and a branch, determining an invalid branch in the power grid; based on invalid branches and a greedy selection mechanism, traversing the bus to screen valid branches, and obtaining a modified branch set; and changing admittance values of branches in the branch set based on a moving target defense strategy, changing a measurement matrix, and realizing low-overhead moving target defense. Compared with the prior art, on the premise that the high detection probability is maintained, the number of branches needing to modify admittance and the calculation time are remarkably reduced, and therefore the system defense cost is reduced.

The present application relates to a kind of new energy fluctuation and data injection attack superimposed cyber-physical fusion hazard evaluation method and system, wherein the method comprises the following steps: constructing new energy fluctuation and data injection attack superimposed attack model, attack model introduces data injection attack vector and the change vector caused by new energy output uncertainty in line flow;Through the calculation of line load security constraint, determine the cyber-physical fusion hazard under new energy fluctuation and data injection attack superposition based on attack model.Compared with the prior art, the present application can determine the worst case by new energy output uncertainty and malicious data attack superposition caused by power grid ontology hazard level, can help operation and maintenance personnel to find the key risk node in power system.

The invention relates to a black box graph injection attack method and device based on a thermonuclear, and aims to solve the problems that the existing graph injection attack (GIA) is insufficient in imperceptibility, limited in application scene and high in calculation cost, and improve the effectiveness and concealment of a graph neural network (GNNs) attack. The method comprises the following steps: firstly, screening candidate nodes based on classification margins and node degrees, and grouping according to most likely misclassification labels; secondly, sampling high-frequency feature dimensions of similar original nodes to generate injection node features consistent with original image feature distribution; then, adopting a two-stage topology construction strategy: constructing a temporary perturbation graph in the first stage, and screening out an injection edge conforming to the topological characteristics of the original graph through node-level thermonuclear signature (HKS) camouflage constraint and graph-level thermonuclear matrix guidance in the second stage; and finally, generating an imperceptible perturbation graph, and reducing the classification performance of the target GNN model on the premise of not exposing attack behaviors. According to the method, the structure or parameters of the target model do not need to be obtained, the method only depends on the black box query and proxy model, the practicability, imperceptibility and calculation efficiency of attacks are remarkably improved, and the method is suitable for GNNs attack testing in safety key scenes such as the Internet of Things.

The invention discloses a novel power system false data injection attack detection method based on CPO-CNN-SE. The method comprises the following steps: preprocessing a measurement test of a novel power system; performing hyper-parameter optimization by using a crown porcupine optimization algorithm (CPO) to improve the precision of the CNN detection model; meanwhile, in consideration of synchronous increase of the FDIA detection speed, a compression excitation (SE) attention mechanism is introduced, a CNN-SE mixed learning model is formed, and the characterization efficiency of the CNN on FDIA spatial features is enhanced; and finally, carrying out binary classification on the FDIA feature data through a full connection layer, a sigmoid classifier and a binary cross entropy loss function. According to the invention, the FDIA detection precision is improved, and the synchronous improvement of the detection speed is realized.

The present application relates to a kind of multi autonomous guide vehicle (AGV) cluster security path tracking control method of resisting false data injection attack, belong to control engineering technical field.For the problem that system stability and safety are threatened by false data injection attack in unreliable network, the present application introduces dynamic event triggering mechanism, only updates control input when necessary, reduces communication burden and energy consumption;Design adaptive state estimator, recover normal system signal from tampered sensor and actuator signal;Adaptive attack compensation mechanism is built, to inhibit the negative influence of attack on system performance.The present application fully considers the application requirement of multi-AGV cluster system in unreliable network environment, by introducing dynamic event triggered control, designing adaptive state estimator and building adaptive attack compensation mechanism, false data injection attack can be effectively resisted, and strong guarantee is provided for multi-AGV cluster system security path tracking control.

The application provides a false data injection attack detection method, a detection terminal and a storage medium, and the method comprises the following steps: establishing a power system state estimation model, determining the contribution degree of each device in the power system based on the power system state estimation model; selecting part of the devices in the power system as to-be-detected devices according to the contribution degree of each device, determining an attack detection parameter of the power system, and determining whether the power system is attacked according to the attack detection parameter. The application selects part of the devices in the power system to construct the attack detection parameter for attack detection, can effectively detect the FDIA attack, and improves the ability of the power CPS to resist malicious network attacks.

A multi-agent system encirclement control method triggered by intermittent dynamic events under hybrid attacks comprises the following steps: S1, constructing a directed communication topology and a Laplacian matrix of a multi-agent system according to an actual task; s2, constructing a multi-agent system model including uncertain disturbance, nonlinear dynamics and false data injection attacks; s3, designing a sliding-mode observer and a double-layer attack detection mechanism for denial of service attack and false data injection attack; s4, designing a fixed time encircling controller of the intermittent dynamic event triggered multi-agent system; s5, the controller designed in the step S4 is used for achieving encircling control within fixed time, and the upper bound of convergence time is obtained; and S6, continuously operating until the encircling control of the multi-agent system is completed. The invention aims to solve the problem of safe, efficient and rapid encircling control when a multi-agent system is subjected to DoS attack, FDI attack and external disturbance at the same time.

The invention provides a false data injection attack identification method based on an adaptive residual weighted PINN. The method comprises the following steps: constructing a physical equation describing a system operation state; a self-adaptive residual weighting network is constructed, and the self-adaptive residual weighting network takes sensor measurement data with noise or attack as input and takes the reconstructed system state as output; training the adaptive residual weighting network by using a gradient descent algorithm; mapping sensor data into a system state meeting physical consistency in real time by using an adaptive residual weighting network; and according to a weight coefficient and physical residual information which are calculated in real time by the self-adaptive residual weighting network, identifying a false data injection attack through a dual-criterion mechanism.

The invention discloses a method for discovering SQL injection attack behaviors based on chaotic parting dimensions, and the method specifically comprises the steps: S1, data collection and feature extraction: obtaining HTTP request data, including SQL query data; extracting the characteristics of the SQL injection attack, wherein the characteristics comprise special characters, SQL keywords and potential malicious structures; s2, calculating a parting dimension; s3, setting a classification dimension threshold value, and if the classification dimension of the selected SQL statement is higher than the threshold value, judging that the SQL statement is an SQL injection attack; and S4, performing anomaly detection and response. The invention mainly relates to the technical field of network security, chaos theory and typing analysis. According to the method, the data is analyzed by using the typing dimension in the chaos theory, the accuracy and efficiency of attack detection are further improved, and nonlinear characteristics and complexity in an attack mode can be revealed, so that the method has advantages in coping with complex attacks.

Aspects of this disclosure relate to systems and methods for preventing injection attacks to large language models (LLMs). Techniques implemented by these systems and methods can include obtaining a query from a user device that is configured to cause an LLM to generate an output outside of a predetermined scope. In response to obtaining the query, systems can execute one or more evaluation models that are configured to identify one or more compatibility issues that can arise if the queries are executed by an LLM. Where a query is identified as having one compatibility issue(s), the systems described herein can either update the query to address the issue. Additionally, or alternatively, the systems described herein can prevent the processing of the query (and processing in furtherance of the injection attack) by an LLM.