{"id":4055,"date":"2019-11-20T10:11:21","date_gmt":"2019-11-20T08:11:21","guid":{"rendered":"https:\/\/blog.eset.ee\/?p=4055"},"modified":"2019-11-20T10:11:41","modified_gmt":"2019-11-20T08:11:41","slug":"evading-machine-learning-detection-in-a-cyber-secure-world","status":"publish","type":"post","link":"https:\/\/blog.eset.ee\/et\/en\/2019\/11\/20\/evading-machine-learning-detection-in-a-cyber-secure-world\/","title":{"rendered":"Evading machine learning detection in a cyber-secure world"},"content":{"rendered":"\n

Return\nof the malware titans<\/strong><\/p>\n\n\n\n

With the announcement of a bypass<\/a>\nof a popular machine learning detection engine earlier this year, many\ndelusions of grandeur of machine learning technology certainly came crashing\ndown. The bypass consisted of a simple appending of \u201chappy strings\u201d pulled from\ngaming software onto a number of malware files \u2013 a procedure known as overlaying<\/em>. Thereby, such evil titans as\nWannaCry<\/a>\nand SamSam<\/a>\nransomware quickly eluded detection.<\/p>\n\n\n\n

A\nnew machine learning evasion competition<\/strong><\/p>\n\n\n\n

Such incidents are salutary reminders of\nthe need for greater awareness of the true state of machine learning in\ncybersecurity \u2013 both its benefits and pitfalls. In this spirit, Jakub Debski,\nESET chief product officer, put his shoulder to the wheel to take on a machine\nlearning static evasion<\/em> competition<\/a>\nhosted by VMRay, Endgame, and MRG-Effitas this past August. Despite the stiff\ncompetition and technical problems in this first-ever hosted competition,\nDebski made ESET proud by achieving a top score with 133 out of 150 points.\n[Note: Placing and score is subject to final revision due to issues with the\ntesting environment. Stay tuned to Endgame\u2019s blog<\/a> for any forthcoming official\nresults.]<\/p>\n\n\n\n

The steps taken were not easy. Altering 50\nmalware samples to avoid detection from three machine learning detection models\n(two MalConv models and one EMBER LightGBM model) while ensuring fully\nfunctioning malware after modification is no mean feat.<\/p>\n\n\n\n

First, Debski discovered that he could\nmanipulate a header entry in the malware samples that was acting as a\nparticularly strong feature for the EMBER classifier. The trick here was to\napply a customized UPX packer and then run a fuzzing script to trick the EMBER\nclassifier about the \u201cbenign-looking\u201d character of the malware. UPX is a common\nenough packer that many machine learning engines would fail to flag.<\/p>\n\n\n\n

The second part of the competition remained\n\u2013 how can an attacker bypass the two MalConv machine learning models? Debski\nappended some very high-scoring \u201chappy strings\u201d on to the malicious samples \u2013\nthe good ol\u2019 overlaying<\/em> trick again \u2013\nwhich ultimately led to the end of the competition.<\/p>\n\n\n\n

You can check out Debski\u2019s play-by-play of\nthe competition here<\/a>.<\/p>\n\n\n\n

Real-world\nattackers upping the ante<\/strong><\/p>\n\n\n\n

To be sure, attackers have much more in\ntheir arsenal than the simple evasion techniques allowed in the competition.\nPerhaps it was \u201cthinking too much as an attacker\u201d that gave other competitors a\nslight head start before Debski put aside his \u201cattacker\u2019s hat,\u201d donned his\n\u201cplayer\u2019s jersey,\u201d and buckled down on gaining the full competition points.\nReal life is messier than competitions, and there are no constraints to an\nattacker\u2019s methods. Just like Debski\u2019s preliminary forays in the competition,\nan attacker would most likely first search for a hole in the security\nenvironment \u2013 think crashing the feature extractor \u2013 as a leverage to entirely\ncircumvent the useful application of any machine learning detections set in\nplace.<\/p>\n\n\n\n

Alternatively, an attacker could also take\na more dynamic approach by using self-extractors and droppers that would easily\nbreeze past any kind of static machine learning models, as employed in this\ncompetition. Since a static analysis is restricted to a point in time, it is unable\nto reveal the nefarious behavior hidden in a time-delayed self-extractor.\nMoreover, when the data is compressed and\/or encrypted inside a self-extractor,\nit is unable to extract anything useful \u2013 it\u2019s all noise. Training the machine\nlearning algorithm to detect all\nself-extractors<\/em> \u2013 which are commonly contained in clean application\ninstallers \u2013 would also cause too many false positives.<\/p>\n\n\n\n

Improve,\nlayer, and protect<\/strong><\/p>\n\n\n\n

Ultimately, what this machine learning\ncompetition demonstrated is the precariousness of betting all your money on one\nhorse \u2013 machine learning. Realizing the wide proliferation of ready-to-use\nmachine learning evasion techniques, ESET places great emphasis on using skilled and experienced malware analysts<\/strong>\nto supplement and ensure that machine learning detection algorithms are not\nleft entirely to their own mysterious machinations.<\/p>\n\n\n\n

That\u2019s why building a machine learning\nengine to detect malware, like ESET\u2019s\nAugur<\/a>, is a continued responsibility on the part of ESET. A machine\nlearning engine needs to be constantly\nupdated<\/strong> as the behaviors of malicious actors become increasingly\nsophisticated and creative. That fresh data is a crucial input for the quality\nof any machine learning detection engine. At ESET, we\u2019ve spent three decades \u201ctinkering\nthe engine\u201d to reduce false positives and misses.<\/p>\n\n\n\n

Another method to improve machine learning\ndetection is to ensure that you are unpacking<\/strong>\ndata before feeding it into a machine learning engine. This protects better\nagainst machine learning evasion techniques via self-extractors, and makes it\npossible to find,<\/strong> emulate, and analyze malware at the\nbehavioral level<\/strong>.<\/p>\n\n\n\n

Each endpoint and device protected by ESET with\nESET LiveGrid\u00ae<\/a> turned on benefits from Augur\u2019s ability\nto analyze emerging threats. Augur also works offline as a lightweight Advanced\nMachine Learning module to protect machines even when they are running without\ninternet access. Enterprise\nclients have Augur at their disposal via ESET Dynamic\nThreat Defense<\/a> (EDTD).<\/p>\n\n\n\n

While machine learning is a great\ntechnology, no business can rely on it alone to stop all adversaries. A robust\nsecurity posture for a business demands multilayered defenses<\/a>,\nlike ESET\u2019s UEFI\nScanner<\/a> and Advanced Memory\nScanner<\/a>, that can protect endpoints from all vectors. Clever adversaries\nmight know how to get past machine learning defenses, but they can\u2019t get far\nwhen other layers are also in place.<\/p>\n","protected":false},"excerpt":{"rendered":"

Return of the malware titans With the announcement of a bypass of a popular machine learning detection engine earlier this year, many delusions of grandeur of machine learning technology certainly came crashing down. The bypass consisted of a simple appending of \u201chappy strings\u201d pulled from gaming software onto a number of malware files \u2013 a<\/p>\n","protected":false},"author":5,"featured_media":4052,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"footnotes":""},"categories":[148],"tags":[],"class_list":["post-4055","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-cybersecurity"],"acf":[],"yoast_head":"\nEvading machine learning detection in a cyber-secure world - ESET Eesti Blogi<\/title>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/blog.eset.ee\/et\/en\/2019\/11\/20\/evading-machine-learning-detection-in-a-cyber-secure-world\/\" \/>\n<meta property=\"og:locale\" content=\"en_US\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"Evading machine learning detection in a cyber-secure world\" \/>\n<meta property=\"og:description\" content=\"Return of the malware titans With the announcement of a bypass of a popular machine learning detection engine earlier this year, many delusions of grandeur of machine learning technology certainly came crashing down. The bypass consisted of a simple appending of \u201chappy strings\u201d pulled from gaming software onto a number of malware files \u2013 a\" \/>\n<meta property=\"og:url\" content=\"https:\/\/blog.eset.ee\/et\/en\/2019\/11\/20\/evading-machine-learning-detection-in-a-cyber-secure-world\/\" \/>\n<meta property=\"og:site_name\" content=\"ESET Eesti Blogi\" \/>\n<meta property=\"article:publisher\" content=\"http:\/\/www.facebook.com\/antiviirus\" \/>\n<meta property=\"article:published_time\" content=\"2019-11-20T08:11:21+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2019-11-20T08:11:41+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/blog.eset.ee\/wp-content\/uploads\/2019\/11\/Machinelearning.png\" \/>\n\t<meta property=\"og:image:width\" content=\"1200\" \/>\n\t<meta property=\"og:image:height\" content=\"628\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/png\" \/>\n<meta name=\"author\" content=\"ESET Blog\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"Written by\" \/>\n\t<meta name=\"twitter:data1\" content=\"ESET Blog\" \/>\n\t<meta name=\"twitter:label2\" content=\"Est. reading time\" \/>\n\t<meta name=\"twitter:data2\" content=\"4 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\\\/\\\/schema.org\",\"@graph\":[{\"@type\":\"Article\",\"@id\":\"https:\\\/\\\/blog.eset.ee\\\/et\\\/en\\\/2019\\\/11\\\/20\\\/evading-machine-learning-detection-in-a-cyber-secure-world\\\/#article\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/blog.eset.ee\\\/et\\\/en\\\/2019\\\/11\\\/20\\\/evading-machine-learning-detection-in-a-cyber-secure-world\\\/\"},\"author\":{\"name\":\"ESET Blog\",\"@id\":\"https:\\\/\\\/blog.eset.ee\\\/et\\\/en\\\/#\\\/schema\\\/person\\\/876cf293277fc0b2ae2f4395fffe4c88\"},\"headline\":\"Evading machine learning detection in a cyber-secure world\",\"datePublished\":\"2019-11-20T08:11:21+00:00\",\"dateModified\":\"2019-11-20T08:11:41+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\\\/\\\/blog.eset.ee\\\/et\\\/en\\\/2019\\\/11\\\/20\\\/evading-machine-learning-detection-in-a-cyber-secure-world\\\/\"},\"wordCount\":893,\"image\":{\"@id\":\"https:\\\/\\\/blog.eset.ee\\\/et\\\/en\\\/2019\\\/11\\\/20\\\/evading-machine-learning-detection-in-a-cyber-secure-world\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/blog.eset.ee\\\/wp-content\\\/uploads\\\/2019\\\/11\\\/Machinelearning.png\",\"articleSection\":[\"cybersecurity\"],\"inLanguage\":\"en-US\",\"copyrightYear\":\"2019\",\"copyrightHolder\":{\"@id\":\"https:\\\/\\\/blog.eset.ee\\\/et\\\/#organization\"}},{\"@type\":\"WebPage\",\"@id\":\"https:\\\/\\\/blog.eset.ee\\\/et\\\/en\\\/2019\\\/11\\\/20\\\/evading-machine-learning-detection-in-a-cyber-secure-world\\\/\",\"url\":\"https:\\\/\\\/blog.eset.ee\\\/et\\\/en\\\/2019\\\/11\\\/20\\\/evading-machine-learning-detection-in-a-cyber-secure-world\\\/\",\"name\":\"Evading machine learning detection in a cyber-secure world - ESET Eesti Blogi\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/blog.eset.ee\\\/et\\\/en\\\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\\\/\\\/blog.eset.ee\\\/et\\\/en\\\/2019\\\/11\\\/20\\\/evading-machine-learning-detection-in-a-cyber-secure-world\\\/#primaryimage\"},\"image\":{\"@id\":\"https:\\\/\\\/blog.eset.ee\\\/et\\\/en\\\/2019\\\/11\\\/20\\\/evading-machine-learning-detection-in-a-cyber-secure-world\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/blog.eset.ee\\\/wp-content\\\/uploads\\\/2019\\\/11\\\/Machinelearning.png\",\"datePublished\":\"2019-11-20T08:11:21+00:00\",\"dateModified\":\"2019-11-20T08:11:41+00:00\",\"author\":{\"@id\":\"https:\\\/\\\/blog.eset.ee\\\/et\\\/en\\\/#\\\/schema\\\/person\\\/876cf293277fc0b2ae2f4395fffe4c88\"},\"breadcrumb\":{\"@id\":\"https:\\\/\\\/blog.eset.ee\\\/et\\\/en\\\/2019\\\/11\\\/20\\\/evading-machine-learning-detection-in-a-cyber-secure-world\\\/#breadcrumb\"},\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\\\/\\\/blog.eset.ee\\\/et\\\/en\\\/2019\\\/11\\\/20\\\/evading-machine-learning-detection-in-a-cyber-secure-world\\\/\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\\\/\\\/blog.eset.ee\\\/et\\\/en\\\/2019\\\/11\\\/20\\\/evading-machine-learning-detection-in-a-cyber-secure-world\\\/#primaryimage\",\"url\":\"https:\\\/\\\/blog.eset.ee\\\/wp-content\\\/uploads\\\/2019\\\/11\\\/Machinelearning.png\",\"contentUrl\":\"https:\\\/\\\/blog.eset.ee\\\/wp-content\\\/uploads\\\/2019\\\/11\\\/Machinelearning.png\",\"width\":1200,\"height\":628},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\\\/\\\/blog.eset.ee\\\/et\\\/en\\\/2019\\\/11\\\/20\\\/evading-machine-learning-detection-in-a-cyber-secure-world\\\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\\\/\\\/blog.eset.ee\\\/et\\\/en\\\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"Evading machine learning detection in a cyber-secure world\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\\\/\\\/blog.eset.ee\\\/et\\\/en\\\/#website\",\"url\":\"https:\\\/\\\/blog.eset.ee\\\/et\\\/en\\\/\",\"name\":\"ESET Eesti Blogi\",\"description\":\"Uudised IT maailmast\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\\\/\\\/blog.eset.ee\\\/et\\\/en\\\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"en-US\"},{\"@type\":\"Person\",\"@id\":\"https:\\\/\\\/blog.eset.ee\\\/et\\\/en\\\/#\\\/schema\\\/person\\\/876cf293277fc0b2ae2f4395fffe4c88\",\"name\":\"ESET Blog\",\"sameAs\":[\"http:\\\/\\\/eset.ee\"],\"url\":\"https:\\\/\\\/blog.eset.ee\\\/et\\\/en\\\/author\\\/allankinsigo\\\/\"},{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\\\/\\\/blog.eset.ee\\\/et\\\/en\\\/2019\\\/11\\\/20\\\/evading-machine-learning-detection-in-a-cyber-secure-world\\\/#local-main-organization-logo\",\"url\":\"\",\"contentUrl\":\"\",\"caption\":\"ESET EESTI\"}]}<\/script>\n<!-- \/ Yoast SEO Premium plugin. -->","yoast_head_json":{"title":"Evading machine learning detection in a cyber-secure world - ESET Eesti Blogi","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/blog.eset.ee\/et\/en\/2019\/11\/20\/evading-machine-learning-detection-in-a-cyber-secure-world\/","og_locale":"en_US","og_type":"article","og_title":"Evading machine learning detection in a cyber-secure world","og_description":"Return of the malware titans With the announcement of a bypass of a popular machine learning detection engine earlier this year, many delusions of grandeur of machine learning technology certainly came crashing down. The bypass consisted of a simple appending of \u201chappy strings\u201d pulled from gaming software onto a number of malware files \u2013 a","og_url":"https:\/\/blog.eset.ee\/et\/en\/2019\/11\/20\/evading-machine-learning-detection-in-a-cyber-secure-world\/","og_site_name":"ESET Eesti Blogi","article_publisher":"http:\/\/www.facebook.com\/antiviirus","article_published_time":"2019-11-20T08:11:21+00:00","article_modified_time":"2019-11-20T08:11:41+00:00","og_image":[{"width":1200,"height":628,"url":"https:\/\/blog.eset.ee\/wp-content\/uploads\/2019\/11\/Machinelearning.png","type":"image\/png"}],"author":"ESET Blog","twitter_card":"summary_large_image","twitter_misc":{"Written by":"ESET Blog","Est. reading time":"4 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/blog.eset.ee\/et\/en\/2019\/11\/20\/evading-machine-learning-detection-in-a-cyber-secure-world\/#article","isPartOf":{"@id":"https:\/\/blog.eset.ee\/et\/en\/2019\/11\/20\/evading-machine-learning-detection-in-a-cyber-secure-world\/"},"author":{"name":"ESET Blog","@id":"https:\/\/blog.eset.ee\/et\/en\/#\/schema\/person\/876cf293277fc0b2ae2f4395fffe4c88"},"headline":"Evading machine learning detection in a cyber-secure world","datePublished":"2019-11-20T08:11:21+00:00","dateModified":"2019-11-20T08:11:41+00:00","mainEntityOfPage":{"@id":"https:\/\/blog.eset.ee\/et\/en\/2019\/11\/20\/evading-machine-learning-detection-in-a-cyber-secure-world\/"},"wordCount":893,"image":{"@id":"https:\/\/blog.eset.ee\/et\/en\/2019\/11\/20\/evading-machine-learning-detection-in-a-cyber-secure-world\/#primaryimage"},"thumbnailUrl":"https:\/\/blog.eset.ee\/wp-content\/uploads\/2019\/11\/Machinelearning.png","articleSection":["cybersecurity"],"inLanguage":"en-US","copyrightYear":"2019","copyrightHolder":{"@id":"https:\/\/blog.eset.ee\/et\/#organization"}},{"@type":"WebPage","@id":"https:\/\/blog.eset.ee\/et\/en\/2019\/11\/20\/evading-machine-learning-detection-in-a-cyber-secure-world\/","url":"https:\/\/blog.eset.ee\/et\/en\/2019\/11\/20\/evading-machine-learning-detection-in-a-cyber-secure-world\/","name":"Evading machine learning detection in a cyber-secure world - ESET Eesti Blogi","isPartOf":{"@id":"https:\/\/blog.eset.ee\/et\/en\/#website"},"primaryImageOfPage":{"@id":"https:\/\/blog.eset.ee\/et\/en\/2019\/11\/20\/evading-machine-learning-detection-in-a-cyber-secure-world\/#primaryimage"},"image":{"@id":"https:\/\/blog.eset.ee\/et\/en\/2019\/11\/20\/evading-machine-learning-detection-in-a-cyber-secure-world\/#primaryimage"},"thumbnailUrl":"https:\/\/blog.eset.ee\/wp-content\/uploads\/2019\/11\/Machinelearning.png","datePublished":"2019-11-20T08:11:21+00:00","dateModified":"2019-11-20T08:11:41+00:00","author":{"@id":"https:\/\/blog.eset.ee\/et\/en\/#\/schema\/person\/876cf293277fc0b2ae2f4395fffe4c88"},"breadcrumb":{"@id":"https:\/\/blog.eset.ee\/et\/en\/2019\/11\/20\/evading-machine-learning-detection-in-a-cyber-secure-world\/#breadcrumb"},"inLanguage":"en-US","potentialAction":[{"@type":"ReadAction","target":["https:\/\/blog.eset.ee\/et\/en\/2019\/11\/20\/evading-machine-learning-detection-in-a-cyber-secure-world\/"]}]},{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/blog.eset.ee\/et\/en\/2019\/11\/20\/evading-machine-learning-detection-in-a-cyber-secure-world\/#primaryimage","url":"https:\/\/blog.eset.ee\/wp-content\/uploads\/2019\/11\/Machinelearning.png","contentUrl":"https:\/\/blog.eset.ee\/wp-content\/uploads\/2019\/11\/Machinelearning.png","width":1200,"height":628},{"@type":"BreadcrumbList","@id":"https:\/\/blog.eset.ee\/et\/en\/2019\/11\/20\/evading-machine-learning-detection-in-a-cyber-secure-world\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/blog.eset.ee\/et\/en\/"},{"@type":"ListItem","position":2,"name":"Evading machine learning detection in a cyber-secure world"}]},{"@type":"WebSite","@id":"https:\/\/blog.eset.ee\/et\/en\/#website","url":"https:\/\/blog.eset.ee\/et\/en\/","name":"ESET Eesti Blogi","description":"Uudised IT maailmast","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/blog.eset.ee\/et\/en\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"en-US"},{"@type":"Person","@id":"https:\/\/blog.eset.ee\/et\/en\/#\/schema\/person\/876cf293277fc0b2ae2f4395fffe4c88","name":"ESET Blog","sameAs":["http:\/\/eset.ee"],"url":"https:\/\/blog.eset.ee\/et\/en\/author\/allankinsigo\/"},{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/blog.eset.ee\/et\/en\/2019\/11\/20\/evading-machine-learning-detection-in-a-cyber-secure-world\/#local-main-organization-logo","url":"","contentUrl":"","caption":"ESET EESTI"}]}},"amp_enabled":true,"_links":{"self":[{"href":"https:\/\/blog.eset.ee\/et\/en\/wp-json\/wp\/v2\/posts\/4055","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/blog.eset.ee\/et\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/blog.eset.ee\/et\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/blog.eset.ee\/et\/en\/wp-json\/wp\/v2\/users\/5"}],"replies":[{"embeddable":true,"href":"https:\/\/blog.eset.ee\/et\/en\/wp-json\/wp\/v2\/comments?post=4055"}],"version-history":[{"count":0,"href":"https:\/\/blog.eset.ee\/et\/en\/wp-json\/wp\/v2\/posts\/4055\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/blog.eset.ee\/et\/en\/wp-json\/wp\/v2\/media\/4052"}],"wp:attachment":[{"href":"https:\/\/blog.eset.ee\/et\/en\/wp-json\/wp\/v2\/media?parent=4055"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/blog.eset.ee\/et\/en\/wp-json\/wp\/v2\/categories?post=4055"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/blog.eset.ee\/et\/en\/wp-json\/wp\/v2\/tags?post=4055"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}