{"id":13844,"date":"2025-09-11T18:30:00","date_gmt":"2025-09-11T16:30:00","guid":{"rendered":"https:\/\/haimagazine.com\/uncategorized\/aegisai-wants-to-stop-phishing-before-you-click\/"},"modified":"2025-09-12T12:09:18","modified_gmt":"2025-09-12T10:09:18","slug":"aegisai-wants-to-stop-phishing-before-you-click","status":"publish","type":"post","link":"https:\/\/haimagazine.com\/en\/ai-news-2\/aegisai-wants-to-stop-phishing-before-you-click\/","title":{"rendered":"\ud83d\udd12 AegisAI wants to stop phishing before you click"},"content":{"rendered":"<p>$4.45 million \u2014 that&#8217;s the average cost of a single data breach caused by <strong>phishing<\/strong> today. According to <strong>IBM<\/strong>, it remains the most common entry point for cybercriminals. But email has long ceased to be just a carrier of extorted logins. Today, email channels <em><strong>ransomware<\/strong><\/em>, <strong>malicious software<\/strong>, and even <strong>sophisticated scams<\/strong> \u2014 from fake invoices to messages impersonating board members. The more personalized and convincing they are, the more effective they become, and those created by AI are increasingly difficult to distinguish from real ones. <strong>According to a CrowdStrike study, phishing generated by large language models achieves a click-through rate of 54%<\/strong>. In comparison, classic emails written by humans only achieves a 12%.<\/p><h4 class=\"wp-block-heading\"><strong>New creators, old mistakes<\/strong><\/h4><p><strong>Cy Khormaee<\/strong> and <strong>Ryan Luo<\/strong>, former <strong>Google<\/strong> managers known for their work on <strong>reCAPTCHA<\/strong>. are entering the game. Their new startup <strong>AegisAI<\/strong> just secured $13 million in funding. Their goal is to approach email security using the same technologies that attackers use today. Khormaee diagnoses the problem frankly: for 30 years, email protection systems haven&#8217;t kept up with threats. Initially, they relied on the sender\u2019s reputation \u2014 today, that\u2019s childishly easy to circumvent. Later came the rule-based filters trained on past incidents. The issue is that today&#8217;s attacks aren&#8217;t repeated \u2014 they&#8217;re unique, dynamic, contextual. By the time a filter learns, it&#8217;s already too late.<\/p><h4 class=\"wp-block-heading\"><strong>Distributed defense, immediate response<\/strong><\/h4><p>Instead of one filter, AegisAI is creating a dispersed network of autonomous agents \u2014 each responsible for a different impact area: links, attachments, QR codes and metadata. Together, they form a dynamic system that analyzes messages in real time and learns from each approach. The models are trained not only on real attacks but also on simulations of adversary behavior. <strong>They recognize patterns that humans find difficult to detect, like slight deviations in syntax, anomalies in traffic between domains or unusual redirect sequences.<\/strong> If a message raises suspicions, it goes to quarantine. If the threat is ambiguous, the system doesn&#8217;t overwhelm the team with alerts, but provides a summary and recommendations. The integration with Microsoft 365 and Google Workspace operates through an API, so setting up protection is a matter of a few hours. <strong>In tests, the number of false alarms dropped by as much as 90%.<\/strong><\/p><h4 class=\"wp-block-heading\"><strong><strong>A shield with processor<\/strong><\/strong><\/h4><p>AegisAI focuses on real-time protection \u2014 not based on blacklists or static rules, but on thoroughly analyzing content, style and communication context. The system aims to recognize subtle deviations from an organization&#8217;s typical language and neutralize threats before they reach the recipient. It&#8217;s not about adding a new layer of filters, but about a smart detection mechanism that learns and adapts on its own. If the project succeeds, AegisAI might not only slow down attackers but also take away their advantage. <strong>But can the partnership with Google protect the world from the weaponry it helped to arm?<\/strong><\/p><p><\/p>","protected":false},"excerpt":{"rendered":"<p>Phishing attacks are becoming more effective and are increasingly written by AI. AegisAI, a new startup from the creators of reCAPTCHA, is preparing a counterattack using the same weapon.<\/p>\n","protected":false},"author":230,"featured_media":13817,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"rank_math_lock_modified_date":false,"footnotes":""},"categories":[813],"tags":[707,715,916,917],"popular":[],"difficulty-level":[36],"ppma_author":[884],"class_list":["post-13844","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ai-news-2","tag-ai-5","tag-artificial-intelligence","tag-cybersecurity-2","tag-phishing-3","difficulty-level-easy"],"acf":[],"authors":[{"term_id":884,"user_id":230,"is_guest":0,"slug":"karolina-ceron","display_name":"Karolina Cero\u0144","avatar_url":"https:\/\/haimagazine.com\/wp-content\/uploads\/2025\/07\/PXL_20250419_110132091.MP4-scaled.jpg","first_name":"Karolina","last_name":"Cero\u0144","user_url":"","job_title":"","description":"Wsp\u00f3\u0142tw\u00f3rczyni newslettera AI Flash, studentka psychologii i pasjonatka sztucznej inteligencji. Interesuj\u0119 si\u0119 wp\u0142ywem nowych technologii na cz\u0142owieka, a w wolnych chwilach eksperymentuj\u0119 z generatywn\u0105 grafik\u0105 w Midjourney."}],"_links":{"self":[{"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/posts\/13844","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/users\/230"}],"replies":[{"embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/comments?post=13844"}],"version-history":[{"count":1,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/posts\/13844\/revisions"}],"predecessor-version":[{"id":13845,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/posts\/13844\/revisions\/13845"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/media\/13817"}],"wp:attachment":[{"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/media?parent=13844"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/categories?post=13844"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/tags?post=13844"},{"taxonomy":"popular","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/popular?post=13844"},{"taxonomy":"difficulty-level","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/difficulty-level?post=13844"},{"taxonomy":"author","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/ppma_author?post=13844"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}