{"id":21124,"date":"2024-07-03T11:54:49","date_gmt":"2024-07-03T18:54:49","guid":{"rendered":"https:\/\/webdev.securin.xyz\/?p=21124"},"modified":"2024-10-22T09:49:28","modified_gmt":"2024-10-22T16:49:28","slug":"bringing-in-the-bom-squad-part-2-ai-ml-libraries-and-the-vulnerabilities-within","status":"publish","type":"post","link":"https:\/\/webdev.securin.xyz\/articles\/bringing-in-the-bom-squad-part-2-ai-ml-libraries-and-the-vulnerabilities-within\/","title":{"rendered":"Bringing in the BoM Squad, Part 2: AI\/ML Libraries and the Vulnerabilities Within"},"content":{"rendered":"\t\t<div data-elementor-type=\"wp-post\" data-elementor-id=\"21124\" class=\"elementor elementor-21124\" data-elementor-post-type=\"post\">\n\t\t\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-10939d4 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"10939d4\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-64ad70c\" data-id=\"64ad70c\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<section class=\"elementor-section elementor-inner-section elementor-element elementor-element-8491e8a elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"8491e8a\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-inner-column elementor-element elementor-element-bedef71\" data-id=\"bedef71\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-f005671 elementor-widget elementor-widget-text-editor\" data-id=\"f005671\" data-element_type=\"widget\" data-settings=\"{&quot;ekit_we_effect_on&quot;:&quot;none&quot;}\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p style=\"text-align: left;\"><span style=\"font-weight: 400;\">One of the most pressing concerns in AI security is the presence of vulnerabilities within AI\/ML libraries. These libraries are the building blocks for developing sophisticated AI models and applications, but can harbor critical security flaws that, if exploited, could have severe consequences, for example:\u00a0<\/span><\/p><ul style=\"text-align: left;\"><li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\"><a href=\"https:\/\/nvd.nist.gov\/vuln\/detail\/CVE-2024-5480\" target=\"_blank\" rel=\"noopener\">CVE-2024-5480<\/a>: Remote Code Execution in Pytorch\u00a0<\/span><\/li><li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\"><a href=\"https:\/\/nvd.nist.gov\/vuln\/detail\/CVE-2024-0520\" target=\"_blank\" rel=\"noopener\">CVE-2024-0520<\/a>: mlflow Arbitrary File Overwrite<\/span><\/li><\/ul><p style=\"text-align: left;\"><span style=\"font-weight: 400;\">Both of these vulnerabilities could create pathways for attackers to compromise AI systems. Understanding and managing these vulnerabilities is essential for maintaining a robust AI security posture.<\/span><\/p><p style=\"text-align: left;\"><span style=\"font-weight: 400;\">We combined our AIBoM research with Securin Vulnerability Intelligence to bring a comprehensive and proactive approach to identifying, analyzing and mitigating vulnerabilities in AI\/ML libraries, enhancing the security posture and resilience of AI-driven systems.\u00a0 Here\u2019s how we did it.<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-e894a3f elementor-widget elementor-widget-image\" data-id=\"e894a3f\" data-element_type=\"widget\" data-settings=\"{&quot;ekit_we_effect_on&quot;:&quot;none&quot;}\" data-widget_type=\"image.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t\t\t\t\t<img fetchpriority=\"high\" decoding=\"async\" width=\"809\" height=\"777\" src=\"https:\/\/webdev.securin.xyz\/wp-content\/uploads\/2024\/07\/AI_ML-Libraries_0.png\" class=\"attachment-full size-full wp-image-21151\" alt=\"\" srcset=\"https:\/\/webdev.securin.xyz\/wp-content\/uploads\/2024\/07\/AI_ML-Libraries_0.png 809w, https:\/\/webdev.securin.xyz\/wp-content\/uploads\/2024\/07\/AI_ML-Libraries_0-300x288.png 300w, https:\/\/webdev.securin.xyz\/wp-content\/uploads\/2024\/07\/AI_ML-Libraries_0-768x738.png 768w\" sizes=\"(max-width: 809px) 100vw, 809px\" \/>\t\t\t\t\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-505f542 elementor-widget elementor-widget-spacer\" data-id=\"505f542\" data-element_type=\"widget\" data-settings=\"{&quot;ekit_we_effect_on&quot;:&quot;none&quot;}\" data-widget_type=\"spacer.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-536d24b elementor-widget elementor-widget-heading\" data-id=\"536d24b\" data-element_type=\"widget\" data-settings=\"{&quot;ekit_we_effect_on&quot;:&quot;none&quot;}\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<h2 class=\"elementor-heading-title elementor-size-default\">A Closer Inspection of AI\/ML Libraries<\/h2>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-c27d6d1 elementor-widget elementor-widget-text-editor\" data-id=\"c27d6d1\" data-element_type=\"widget\" data-settings=\"{&quot;ekit_we_effect_on&quot;:&quot;none&quot;}\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">In our <\/span><a href=\"https:\/\/www.securin.io\/articles\/bringing-in-the-bom-squad-defining-and-generating-ai-bill-of-materials\/\" target=\"_blank\" rel=\"noopener\"><span style=\"font-weight: 400;\">previous blog<\/span><\/a><span style=\"font-weight: 400;\">, we detailed the generation and construction of an AI Bill of Materials (AIBoM) across ~500k Models in the Hugging Face repository. The programming libraries and packages that form the foundation of AI and ML development are a crucial element of our AIBoMs, providing essential tools and functionalities for building, training and deploying intelligent models and applications. We identified 3000+ libraries and ranked them by their frequency, noting expected guests such as:<\/span><\/p><ul><li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400; color: #000000;\"><strong>transformers:<\/strong> 295,242 models<\/span><\/li><li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400; color: #000000;\"><strong>pytorch:<\/strong> 175,423 models<\/span><\/li><li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400; color: #000000;\"><strong>safetensors:<\/strong> 146,254 models<\/span><\/li><li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400; color: #000000;\"><strong>tensorboard:<\/strong> 90,142 models<\/span><\/li><li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400; color: #000000;\"><strong>diffusers:<\/strong> 25,049 models<\/span><\/li><\/ul>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-5d8b66e elementor-widget elementor-widget-spacer\" data-id=\"5d8b66e\" data-element_type=\"widget\" data-settings=\"{&quot;ekit_we_effect_on&quot;:&quot;none&quot;}\" data-widget_type=\"spacer.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-b7f3952 elementor-widget elementor-widget-heading\" data-id=\"b7f3952\" data-element_type=\"widget\" data-settings=\"{&quot;ekit_we_effect_on&quot;:&quot;none&quot;}\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<h2 class=\"elementor-heading-title elementor-size-default\">Apples &amp; Oranges: Categorizing AI\/ML Libraries<\/h2>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-73e6a71 elementor-widget elementor-widget-text-editor\" data-id=\"73e6a71\" data-element_type=\"widget\" data-settings=\"{&quot;ekit_we_effect_on&quot;:&quot;none&quot;}\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">Categorizing AI\/ML libraries helps streamline the selection process based on specific needs and functionalities. We could clearly see the necessity of differentiating Data Manipulation and Analysis libraries like Pandas and NumPy vs Deep Learning frameworks such as TensorFlow and PyTorch. Our approach therefore categorized the libraries into 14 buckets that enhance efficiency, clarity and specialization in AI and ML development tasks, ensuring the right place of identification within the supply chain:<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-inner-section elementor-element elementor-element-397e4a2 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"397e4a2\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-50 elementor-inner-column elementor-element elementor-element-154fd78\" data-id=\"154fd78\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-ad5408a elementor-widget elementor-widget-image\" data-id=\"ad5408a\" data-element_type=\"widget\" data-settings=\"{&quot;ekit_we_effect_on&quot;:&quot;none&quot;}\" data-widget_type=\"image.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t\t\t\t\t<img decoding=\"async\" width=\"3938\" height=\"3705\" src=\"https:\/\/webdev.securin.xyz\/wp-content\/uploads\/2024\/07\/Categorized-AI_ML-Libraries.png\" class=\"attachment-full size-full wp-image-21143\" alt=\"\" srcset=\"https:\/\/webdev.securin.xyz\/wp-content\/uploads\/2024\/07\/Categorized-AI_ML-Libraries.png 3938w, https:\/\/webdev.securin.xyz\/wp-content\/uploads\/2024\/07\/Categorized-AI_ML-Libraries-300x282.png 300w, https:\/\/webdev.securin.xyz\/wp-content\/uploads\/2024\/07\/Categorized-AI_ML-Libraries-1024x963.png 1024w, https:\/\/webdev.securin.xyz\/wp-content\/uploads\/2024\/07\/Categorized-AI_ML-Libraries-768x723.png 768w, https:\/\/webdev.securin.xyz\/wp-content\/uploads\/2024\/07\/Categorized-AI_ML-Libraries-1536x1445.png 1536w, https:\/\/webdev.securin.xyz\/wp-content\/uploads\/2024\/07\/Categorized-AI_ML-Libraries-2048x1927.png 2048w\" sizes=\"(max-width: 3938px) 100vw, 3938px\" \/>\t\t\t\t\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t<div class=\"elementor-column elementor-col-50 elementor-inner-column elementor-element elementor-element-ae868a0\" data-id=\"ae868a0\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-8b12d00 elementor-widget elementor-widget-text-editor\" data-id=\"8b12d00\" data-element_type=\"widget\" data-settings=\"{&quot;ekit_we_effect_on&quot;:&quot;none&quot;}\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<ul><li><span style=\"font-weight: 400;\">Data Manipulation and Analysis &#8211; 106<\/span><\/li><li><span style=\"font-weight: 400;\">Machine Learning &#8211; 93<\/span><\/li><li><span style=\"font-weight: 400;\">Utilities &#8211; 73<\/span><\/li><li><span style=\"font-weight: 400;\">Deep Learning &#8211; 66<\/span><\/li><li><span style=\"font-weight: 400;\">Natural Language Processing (NLP) &#8211; 58<\/span><\/li><li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">Data Visualization &#8211; 53<\/span><\/li><li><span style=\"font-weight: 400;\">Model Deployment and Serving &#8211; 43<\/span><\/li><li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">Computer Vision &#8211; 41<\/span><\/li><li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">Data Collection and Web Scraping &#8211; 39<\/span><\/li><li><span style=\"font-weight: 400;\">Unknown &#8211; 39<\/span><\/li><li><span style=\"font-weight: 400;\">Reinforcement Learning &#8211; 29<\/span><\/li><li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">Data Storage and Databases &#8211; 27<\/span><\/li><li><span style=\"font-weight: 400;\">Explainability and Interpretability &#8211; 18<\/span><\/li><li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">Optimization and Hyperparameter Tuning &#8211; 12<\/span><\/li><\/ul>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<div class=\"elementor-element elementor-element-1f941bf elementor-widget elementor-widget-spacer\" data-id=\"1f941bf\" data-element_type=\"widget\" data-settings=\"{&quot;ekit_we_effect_on&quot;:&quot;none&quot;}\" data-widget_type=\"spacer.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-15dd5c7 elementor-widget elementor-widget-heading\" data-id=\"15dd5c7\" data-element_type=\"widget\" data-settings=\"{&quot;ekit_we_effect_on&quot;:&quot;none&quot;}\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<h2 class=\"elementor-heading-title elementor-size-default\">Navigating Critical Risks in the Supply Chain<\/h2>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-e40193b elementor-widget elementor-widget-text-editor\" data-id=\"e40193b\" data-element_type=\"widget\" data-settings=\"{&quot;ekit_we_effect_on&quot;:&quot;none&quot;}\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">On to the forefront of supply chain security:\u00a0 our exploration navigates the landscape of potential risks posed by vulnerabilities within these critical frameworks. By uncovering vulnerabilities and their implications, we illuminate the path toward fortifying AI systems with robust defenses and proactive security measures.<\/span><\/p><p><span style=\"font-weight: 400;\">We analyzed the Top 300 python AI\/ML Libraries, leveraging Securin\u2019s Vulnerability Intelligence to map 838 known CVEs for each unique package version.<\/span><\/p><p><span style=\"font-weight: 400;\">With 426 CVEs, TensorFlow takes the first spot in our AI\/ML python libraries ranked by number of direct vulnerabilities. It\u2019s interesting to note the mix of top affected libraries with Deep Learning packages such as TensorFlow, Data Manipulation and Analysis libraries like NumPy as well as utilities like Django making up the Top 10.<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-62a33ab elementor-widget elementor-widget-image\" data-id=\"62a33ab\" data-element_type=\"widget\" data-settings=\"{&quot;ekit_we_effect_on&quot;:&quot;none&quot;}\" data-widget_type=\"image.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t\t\t\t\t<img decoding=\"async\" width=\"1215\" height=\"728\" src=\"https:\/\/webdev.securin.xyz\/wp-content\/uploads\/2024\/07\/Top-AI_ML-Libraries-by-Count-of-CVEs.jpg\" class=\"attachment-full size-full wp-image-21127\" alt=\"\" srcset=\"https:\/\/webdev.securin.xyz\/wp-content\/uploads\/2024\/07\/Top-AI_ML-Libraries-by-Count-of-CVEs.jpg 1215w, https:\/\/webdev.securin.xyz\/wp-content\/uploads\/2024\/07\/Top-AI_ML-Libraries-by-Count-of-CVEs-300x180.jpg 300w, https:\/\/webdev.securin.xyz\/wp-content\/uploads\/2024\/07\/Top-AI_ML-Libraries-by-Count-of-CVEs-1024x614.jpg 1024w, https:\/\/webdev.securin.xyz\/wp-content\/uploads\/2024\/07\/Top-AI_ML-Libraries-by-Count-of-CVEs-768x460.jpg 768w\" sizes=\"(max-width: 1215px) 100vw, 1215px\" \/>\t\t\t\t\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-1d2f082 elementor-widget elementor-widget-spacer\" data-id=\"1d2f082\" data-element_type=\"widget\" data-settings=\"{&quot;ekit_we_effect_on&quot;:&quot;none&quot;}\" data-widget_type=\"spacer.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<section class=\"elementor-section elementor-inner-section elementor-element elementor-element-fa9c8b9 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"fa9c8b9\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-50 elementor-inner-column elementor-element elementor-element-f1d8f76\" data-id=\"f1d8f76\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-4353271 elementor-widget elementor-widget-text-editor\" data-id=\"4353271\" data-element_type=\"widget\" data-settings=\"{&quot;ekit_we_effect_on&quot;:&quot;none&quot;}\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">The threats associated with the vulnerabilities are concerning, with the number of weaponized CVEs (122) and the chatter around them in deep\/dark web forums (353) indicating a growing interest from cybercriminals in exploiting them.<\/span><\/p><p><span style=\"font-weight: 400;\">One of these, <a href=\"https:\/\/nvd.nist.gov\/vuln\/detail\/CVE-2023-4863\" target=\"_blank\" rel=\"noopener\">CVE-2023-4863<\/a>: Heap buffer overflow in libwebp in Google Chrome which has been exploited in the wild, also impacts the pillow library.<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t<div class=\"elementor-column elementor-col-50 elementor-inner-column elementor-element elementor-element-ec681c4\" data-id=\"ec681c4\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-7f1a5f7 elementor-widget elementor-widget-image\" data-id=\"7f1a5f7\" data-element_type=\"widget\" data-settings=\"{&quot;ekit_we_effect_on&quot;:&quot;none&quot;}\" data-widget_type=\"image.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t\t\t\t\t<img loading=\"lazy\" decoding=\"async\" width=\"3071\" height=\"2550\" src=\"https:\/\/webdev.securin.xyz\/wp-content\/uploads\/2024\/07\/Vulnerability-Intelligence-AI_ML-Library-CVE2.png\" class=\"attachment-full size-full wp-image-21144\" alt=\"\" srcset=\"https:\/\/webdev.securin.xyz\/wp-content\/uploads\/2024\/07\/Vulnerability-Intelligence-AI_ML-Library-CVE2.png 3071w, https:\/\/webdev.securin.xyz\/wp-content\/uploads\/2024\/07\/Vulnerability-Intelligence-AI_ML-Library-CVE2-300x249.png 300w, https:\/\/webdev.securin.xyz\/wp-content\/uploads\/2024\/07\/Vulnerability-Intelligence-AI_ML-Library-CVE2-1024x850.png 1024w, https:\/\/webdev.securin.xyz\/wp-content\/uploads\/2024\/07\/Vulnerability-Intelligence-AI_ML-Library-CVE2-768x638.png 768w, https:\/\/webdev.securin.xyz\/wp-content\/uploads\/2024\/07\/Vulnerability-Intelligence-AI_ML-Library-CVE2-1536x1275.png 1536w, https:\/\/webdev.securin.xyz\/wp-content\/uploads\/2024\/07\/Vulnerability-Intelligence-AI_ML-Library-CVE2-2048x1701.png 2048w\" sizes=\"(max-width: 3071px) 100vw, 3071px\" \/>\t\t\t\t\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<div class=\"elementor-element elementor-element-0133cee elementor-widget elementor-widget-spacer\" data-id=\"0133cee\" data-element_type=\"widget\" data-settings=\"{&quot;ekit_we_effect_on&quot;:&quot;none&quot;}\" data-widget_type=\"spacer.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-7364627 elementor-widget elementor-widget-text-editor\" data-id=\"7364627\" data-element_type=\"widget\" data-settings=\"{&quot;ekit_we_effect_on&quot;:&quot;none&quot;}\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">Let\u2019s go another layer deep, drilling down to the weaknesses that foster vulnerabilities within AI\/ML Libraries. Twelve of the MITRE Top 25 Weaknesses are present in the top 25 weaknesses across AI\/ML Library CVEs. Some standouts are:<\/span><\/p><ul><li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\"><a href=\"https:\/\/cwe.mitre.org\/data\/definitions\/369.html\" target=\"_blank\" rel=\"noopener\">CWE-369<\/a>: Divide By Zero &#8211; 51 CVEs<\/span><\/li><li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\"><a href=\"https:\/\/cwe.mitre.org\/data\/definitions\/824.html\" target=\"_blank\" rel=\"noopener\">CWE-824<\/a>: Access of Uninitialized Pointer &#8211; 14 CVEs<\/span><\/li><li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\"><a href=\"https:\/\/cwe.mitre.org\/data\/definitions\/1333.html\" target=\"_blank\" rel=\"noopener\">CWE-1333<\/a>: Inefficient Regular Expression Complexity &#8211; 13 CVEs<\/span><\/li><\/ul><p><span style=\"font-weight: 400;\">These weaknesses are rare occurrences in the larger CVE dataset but become more prominent within AI\/ML frameworks.\u00a0 You won\u2019t see a Software Engineer leave behind a \u201cDivide by Zero\u201d bug! That\u2019s why Data Scientists and Engineers are so different! .<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-a5ff8eb elementor-widget elementor-widget-image\" data-id=\"a5ff8eb\" data-element_type=\"widget\" data-settings=\"{&quot;ekit_we_effect_on&quot;:&quot;none&quot;}\" data-widget_type=\"image.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t\t\t\t\t<img loading=\"lazy\" decoding=\"async\" width=\"1206\" height=\"977\" src=\"https:\/\/webdev.securin.xyz\/wp-content\/uploads\/2024\/07\/Top-25-Weaknesses-Across-AI_ML-Library-CVEs.jpg\" class=\"attachment-full size-full wp-image-21129\" alt=\"\" srcset=\"https:\/\/webdev.securin.xyz\/wp-content\/uploads\/2024\/07\/Top-25-Weaknesses-Across-AI_ML-Library-CVEs.jpg 1206w, https:\/\/webdev.securin.xyz\/wp-content\/uploads\/2024\/07\/Top-25-Weaknesses-Across-AI_ML-Library-CVEs-300x243.jpg 300w, https:\/\/webdev.securin.xyz\/wp-content\/uploads\/2024\/07\/Top-25-Weaknesses-Across-AI_ML-Library-CVEs-1024x830.jpg 1024w, https:\/\/webdev.securin.xyz\/wp-content\/uploads\/2024\/07\/Top-25-Weaknesses-Across-AI_ML-Library-CVEs-768x622.jpg 768w\" sizes=\"(max-width: 1206px) 100vw, 1206px\" \/>\t\t\t\t\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-cdd2be5 elementor-widget elementor-widget-spacer\" data-id=\"cdd2be5\" data-element_type=\"widget\" data-settings=\"{&quot;ekit_we_effect_on&quot;:&quot;none&quot;}\" data-widget_type=\"spacer.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-c825668 elementor-widget elementor-widget-heading\" data-id=\"c825668\" data-element_type=\"widget\" data-settings=\"{&quot;ekit_we_effect_on&quot;:&quot;none&quot;}\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<h2 class=\"elementor-heading-title elementor-size-default\">Operationalizing within the AI\/ML Supply Chain<\/h2>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-0a3924a elementor-widget elementor-widget-text-editor\" data-id=\"0a3924a\" data-element_type=\"widget\" data-settings=\"{&quot;ekit_we_effect_on&quot;:&quot;none&quot;}\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">So now we have the models, their AIBoMs, and visibility across the AI\/ML libraries by analyzing each unique deployment. For example, consider the <\/span><a href=\"https:\/\/github.com\/mistralai\/mistral-inference\" target=\"_blank\" rel=\"noopener\"><span style=\"font-weight: 400;\">mistral-inference<\/span><\/a><span style=\"font-weight: 400;\"> repository to run mistral 7B, 8x7B and 8x22B models. The repository\u2019s poetry consists of minimal package-versions such as numpy (&gt;=1.21.6) and tensorflow (&gt;=2.11.0).\u00a0<\/span><\/p><p><span style=\"font-weight: 400;\">Considering a deployment of mistralai\/Mistral-7B-Instruct-v0.2 with minimal package requirements, we can see the following AIBoM Vulnerabilities (note that the package versions are under the control of the developers and not the LLM provider):<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-01fd20c elementor-widget elementor-widget-image\" data-id=\"01fd20c\" data-element_type=\"widget\" data-settings=\"{&quot;ekit_we_effect_on&quot;:&quot;none&quot;}\" data-widget_type=\"image.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t\t\t\t\t<img loading=\"lazy\" decoding=\"async\" width=\"1373\" height=\"816\" src=\"https:\/\/webdev.securin.xyz\/wp-content\/uploads\/2024\/07\/AI_ML-Libraries-Analyzing-Deployment.png\" class=\"attachment-full size-full wp-image-21150\" alt=\"\" srcset=\"https:\/\/webdev.securin.xyz\/wp-content\/uploads\/2024\/07\/AI_ML-Libraries-Analyzing-Deployment.png 1373w, https:\/\/webdev.securin.xyz\/wp-content\/uploads\/2024\/07\/AI_ML-Libraries-Analyzing-Deployment-300x178.png 300w, https:\/\/webdev.securin.xyz\/wp-content\/uploads\/2024\/07\/AI_ML-Libraries-Analyzing-Deployment-1024x609.png 1024w, https:\/\/webdev.securin.xyz\/wp-content\/uploads\/2024\/07\/AI_ML-Libraries-Analyzing-Deployment-768x456.png 768w\" sizes=\"(max-width: 1373px) 100vw, 1373px\" \/>\t\t\t\t\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-c44ea77 elementor-widget elementor-widget-text-editor\" data-id=\"c44ea77\" data-element_type=\"widget\" data-settings=\"{&quot;ekit_we_effect_on&quot;:&quot;none&quot;}\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">What\u2019s next? We\u2019re glad you asked\u2026<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-b0cad95 elementor-widget elementor-widget-spacer\" data-id=\"b0cad95\" data-element_type=\"widget\" data-settings=\"{&quot;ekit_we_effect_on&quot;:&quot;none&quot;}\" data-widget_type=\"spacer.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-ddd87ad elementor-widget elementor-widget-heading\" data-id=\"ddd87ad\" data-element_type=\"widget\" data-settings=\"{&quot;ekit_we_effect_on&quot;:&quot;none&quot;}\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<h2 class=\"elementor-heading-title elementor-size-default\">Frameworks, Vulnerabilities and AIBoM Components<\/h2>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-1bfe7b8 elementor-widget elementor-widget-text-editor\" data-id=\"1bfe7b8\" data-element_type=\"widget\" data-settings=\"{&quot;ekit_we_effect_on&quot;:&quot;none&quot;}\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">Part three of this series will explore the AI Attack Surface: MITRE ATLAS, OWASP Top 10 for LLMs &amp; ML and how it all fits over the AI Bill of Materials. Our research will provide a comprehensive AI Attack Surface overview, neatly tying together the frameworks, vulnerabilities and AIBoM components.<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-70b482c elementor-widget elementor-widget-text-editor\" data-id=\"70b482c\" data-element_type=\"widget\" data-settings=\"{&quot;ekit_we_effect_on&quot;:&quot;none&quot;}\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><strong>Part 3 Coming Soon:<\/strong><\/p><p><b>The AI Attack Surface &#8211; <\/b><span style=\"font-weight: 400;\">MITRE ATLAS, OWASP Top 10 for LLMs &amp; ML \u2026 how does it all fit over the AI Bill of Materials, our research provides a comprehensive AI Attack Surface overview neatly tying together the frameworks, vulnerabilities and AIBoM components.<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-cc29174 elementor-widget elementor-widget-text-editor\" data-id=\"cc29174\" data-element_type=\"widget\" data-settings=\"{&quot;ekit_we_effect_on&quot;:&quot;none&quot;}\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><a href=\"https:\/\/www.securin.io\/articles\/bringing-in-the-bom-squad-defining-and-generating-ai-bill-of-materials\/\" target=\"_blank\" rel=\"noopener\">Go Back and Read Part 1:<\/a><\/p><p><b>Defining and Generating AI Bill of Materials &#8211; <\/b>AI tools are revolutionizing the tech supply chain. AIBoMs will help us to evolve securely and responsibly. Here\u2019s what you need to know.<\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-b84b971 elementor-widget elementor-widget-spacer\" data-id=\"b84b971\" data-element_type=\"widget\" data-settings=\"{&quot;ekit_we_effect_on&quot;:&quot;none&quot;}\" data-widget_type=\"spacer.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<div class=\"elementor-spacer\">\n\t\t\t<div class=\"elementor-spacer-inner\"><\/div>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<section class=\"elementor-section elementor-inner-section elementor-element elementor-element-30d6fd1 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"30d6fd1\" data-element_type=\"section\" data-settings=\"{&quot;background_background&quot;:&quot;classic&quot;}\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-inner-column elementor-element elementor-element-c96eb62\" data-id=\"c96eb62\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-1a3c486 elementor-widget elementor-widget-text-editor\" data-id=\"1a3c486\" data-element_type=\"widget\" data-settings=\"{&quot;ekit_we_effect_on&quot;:&quot;none&quot;}\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<h3 dir=\"ltr\" style=\"text-align: center;\">Learn more about AI and ML from <a href=\"https:\/\/securin.io\/contact-us\" target=\"_blank\" rel=\"noopener\">our experts<\/a> to strengthen your cybersecurity posture.<\/h3>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<\/div>\n\t\t","protected":false},"excerpt":{"rendered":"<p>Vulnerabilities in AI\/ML libraries are among the most pressing concerns in AI security. Here\u2019s what you need to know.<\/p>\n","protected":false},"author":1,"featured_media":21159,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"content-type":"","footnotes":""},"categories":[631],"tags":[],"acf":[],"aioseo_notices":[],"_links":{"self":[{"href":"https:\/\/webdev.securin.xyz\/wp-json\/wp\/v2\/posts\/21124"}],"collection":[{"href":"https:\/\/webdev.securin.xyz\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/webdev.securin.xyz\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/webdev.securin.xyz\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/webdev.securin.xyz\/wp-json\/wp\/v2\/comments?post=21124"}],"version-history":[{"count":21,"href":"https:\/\/webdev.securin.xyz\/wp-json\/wp\/v2\/posts\/21124\/revisions"}],"predecessor-version":[{"id":21160,"href":"https:\/\/webdev.securin.xyz\/wp-json\/wp\/v2\/posts\/21124\/revisions\/21160"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/webdev.securin.xyz\/wp-json\/wp\/v2\/media\/21159"}],"wp:attachment":[{"href":"https:\/\/webdev.securin.xyz\/wp-json\/wp\/v2\/media?parent=21124"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/webdev.securin.xyz\/wp-json\/wp\/v2\/categories?post=21124"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/webdev.securin.xyz\/wp-json\/wp\/v2\/tags?post=21124"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}