{"id":148425,"date":"2025-09-11T07:26:07","date_gmt":"2025-09-11T07:26:07","guid":{"rendered":"https:\/\/www.newsbeep.com\/us\/148425\/"},"modified":"2025-09-11T07:26:07","modified_gmt":"2025-09-11T07:26:07","slug":"ai-prompt-injection-gets-real-with-macros-the-latest-hidden-threat","status":"publish","type":"post","link":"https:\/\/www.newsbeep.com\/us\/148425\/","title":{"rendered":"AI prompt injection gets real \u2014 with macros the latest hidden threat"},"content":{"rendered":"<p>\u201cAttackers conceal instructions via ultra-small fonts, background-matched text, ASCII smuggling using Unicode Tags, macros that inject payloads at parsing time, and even file metadata (e.g., DOCX custom properties, PDF\/XMP, EXIF),\u201d Grano\u0161a explained. \u201cThese vectors evade human review yet are fully parsed and executed by LLMs, enabling indirect prompt injection.\u201d<\/p>\n<p>Countermeasures<\/p>\n<p>Justin Endres, head of data security at cybersecurity vendor Seclore, argued that security leaders can\u2019t rely on legacy tools alone to defend against malicious prompts that turn \u201ceveryday files into Trojan horses for AI systems.\u201d<\/p>\n<p>\u201c[Security leaders] need layered defenses that sanitize content before it ever reaches an AI parser, enforce strict guardrails around model inputs, and keep humans in the loop for critical workflows,\u201d Endres advised. \u201cOtherwise, attackers will be the ones writing the prompts that shape your AI\u2019s behavior.\u201d<\/p>\n","protected":false},"excerpt":{"rendered":"\u201cAttackers conceal instructions via ultra-small fonts, background-matched text, ASCII smuggling using Unicode Tags, macros that inject payloads at&hellip;\n","protected":false},"author":2,"featured_media":148426,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[45],"tags":[182,181,507,74],"class_list":{"0":"post-148425","1":"post","2":"type-post","3":"status-publish","4":"format-standard","5":"has-post-thumbnail","7":"category-artificial-intelligence","8":"tag-ai","9":"tag-artificial-intelligence","10":"tag-artificialintelligence","11":"tag-technology"},"_links":{"self":[{"href":"https:\/\/www.newsbeep.com\/us\/wp-json\/wp\/v2\/posts\/148425","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.newsbeep.com\/us\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.newsbeep.com\/us\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/us\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/us\/wp-json\/wp\/v2\/comments?post=148425"}],"version-history":[{"count":0,"href":"https:\/\/www.newsbeep.com\/us\/wp-json\/wp\/v2\/posts\/148425\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/us\/wp-json\/wp\/v2\/media\/148426"}],"wp:attachment":[{"href":"https:\/\/www.newsbeep.com\/us\/wp-json\/wp\/v2\/media?parent=148425"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.newsbeep.com\/us\/wp-json\/wp\/v2\/categories?post=148425"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.newsbeep.com\/us\/wp-json\/wp\/v2\/tags?post=148425"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}