Detections
Explore public detection logic contributed by the community across SIEM and rule languages.
379 detections
Filters
Last updated
All Time
Detection languages
170
71
34
34
18
Contributors
89
76
55
21
16
Categories
283
178
76
67
52
Platforms
222
84
83
61
34
Products / Services
92
66
57
37
33
MITRE Techniques
18,123
15,639
12,581
8,394
6,215
CVEs
5
1
1
1
1
IDS Classtypes
23
6
3
1
1
IDS Protocols
13
11
4
3
2
Detects an anomalous volume of file access or synchronization activity on cloud document platforms (Microsoft SharePoint, OneDrive, or Google Drive) that occurs following a high-severity risky sign-in event. The rule specifically flags potential exfiltration patterns such as large volumes of data transfer or numerous file accesses, especially if associated with unmanaged devices or connections to known personal/external file-sharing services.
Detects DNS queries to major public LLM API endpoints (OpenAI, Anthropic, Google Generative Language). This may indicate potential use of these services as a command and control (C2) channel or for data exfiltration by insiders or compromised systems, which may bypass traditional security controls.
Detects potential exfiltration of sensitive DevOps credentials or configuration files (such as AWS credentials, SSH keys, Kubernetes configs, or cryptocurrency wallets) by monitoring for file access events followed by outbound network connections to known exfiltration channels like Slack or specific C2 infrastructure within a 30-minute window.
Detects large HTTP POST requests (exceeding 100KB) directed to public LLM API endpoints such as OpenAI or Anthropic. This behavior may indicate potential bulk data exfiltration or unauthorized sharing of sensitive data with external AI providers.
Detects instances where a non-browser process initiates connections to commercial LLM provider APIs (DeepSeek, OpenRouter, Mistral) followed by a connection to Discord CDN/Webhook endpoints within a 15-minute window. This behavior is indicative of a process acting as an autonomous C2 agent that exfiltrates data after receiving instructions or processing information via an LLM.
Detects anomalous DNS queries that utilize long labels or non-standard record types (TXT, NULL, CNAME) which are frequently associated with DNS tunneling and command-and-control communication. The rule flags endpoints with a high volume of these suspicious requests or exceptionally long domain labels.
Detects the creation or exposure of public repositories, releases, or gists under a personal GitHub account that utilize naming conventions associated with the 'PixelLeak' workaround. This technique allows adversaries to make previously private screenshots, recordings, or internal assets publicly accessible by hosting them in public personal repositories or releases.
Detects the creation or exposure of public repositories, releases, or gists under a personal GitHub account that utilize naming conventions associated with the 'PixelLeak' workaround. This technique allows adversaries to make previously private screenshots, recordings, or internal assets publicly accessible by hosting them in public personal repositories or releases.
Detects the creation or exposure of public repositories, releases, or gists under a personal GitHub account that utilize naming conventions associated with the 'PixelLeak' workaround. This technique allows adversaries to make previously private screenshots, recordings, or internal assets publicly accessible by hosting them in public personal repositories or releases.
This rule detects potentially malicious network activity by monitoring for specific query parameters ('bmodule', 'smodule', 'lmodule', 'task', 'upload.php') in web requests or communication with a specific known-malicious IP address ('62.60.226.50'). This pattern is characteristic of C2 (Command and Control) traffic or unauthorized file staging/upload operations.
Detects a sequence of potentially malicious activity involving the bulk reading or downloading of sensitive AI-related artifacts (model files, datasets), followed by the staging of these files into a compressed archive, and finally an outbound network transfer to an external IP address. This behavior pattern is indicative of data exfiltration of intellectual property.
Detects a suspicious sequence of events where a host performs multiple reconnaissance-style probes (health checks, documentation, or OpenAPI definitions) followed within 10 minutes by access to sensitive artifact-related URLs on the same remote host. This pattern is indicative of an attacker profiling a target server for metadata and then proceeding to exfiltrate files or sensitive artifacts.
Detects a suspicious sequence of events where a host performs multiple reconnaissance-style probes (health checks, documentation, or OpenAPI definitions) followed within 10 minutes by access to sensitive artifact-related URLs on the same remote host. This pattern is indicative of an attacker profiling a target server for metadata and then proceeding to exfiltrate files or sensitive artifacts.
This rule monitors the behavior of AI agent orchestrators to detect potentially malicious activities, specifically focusing on unauthorized tool usage, deep tool invocation chains, and potential data exfiltration. It triggers when an agent invokes a tool outside of its authorized scope, performs an excessively deep sequence of tool calls, or accesses sensitive data followed by an outbound connection to an external endpoint, which is consistent with the MITRE ATLAS AML.T0053 technique.
Detects instances where a non-browser process initiates connections to commercial LLM provider APIs (DeepSeek, OpenRouter, Mistral) followed by a connection to Discord CDN/Webhook endpoints within a 15-minute window. This behavior is indicative of a process acting as an autonomous C2 agent that exfiltrates data after receiving instructions or processing information via an LLM.
Detects anomalous outbound HTTP POST requests containing sensitive file extensions or keywords (e.g., .docx, password, credentials) triggered by Python-based clients, followed by a potential exfiltration beacon response containing JSON file path data.
Detects an endpoint initiating outbound network connections to three or more distinct LLM provider APIs (DeepSeek, Qwen, Mistral, Gemini) within a short timeframe (15-minute windows). This behavior aligns with the multi-provider voting mechanism used by the CLOSEDQUORUM post-compromise framework to autonomously determine subsequent actions using diverse generative AI models.
Detects anomalous, unauthorized access to machine learning model artifacts (e.g., .onnx, .pt, .h5) within object storage or model registries. The rule identifies suspicious patterns including high-volume file reads by non-ML team identities, lack of multi-factor authentication, or access from geolocations outside approved operating regions, indicative of potential model exfiltration or intellectual property theft.
Detects instances where a non-browser process initiates connections to commercial LLM provider APIs (DeepSeek, OpenRouter, Mistral) followed by a connection to Discord CDN/Webhook endpoints within a 15-minute window. This behavior is indicative of a process acting as an autonomous C2 agent that exfiltrates data after receiving instructions or processing information via an LLM.
Detects instances where a non-browser process initiates connections to commercial LLM provider APIs (DeepSeek, OpenRouter, Mistral) followed by a connection to Discord CDN/Webhook endpoints within a 15-minute window. This behavior is indicative of a process acting as an autonomous C2 agent that exfiltrates data after receiving instructions or processing information via an LLM.
Detects instances where a non-browser process initiates connections to commercial LLM provider APIs (DeepSeek, OpenRouter, Mistral) followed by a connection to Discord CDN/Webhook endpoints within a 15-minute window. This behavior is indicative of a process acting as an autonomous C2 agent that exfiltrates data after receiving instructions or processing information via an LLM.
Page 3 of 19