{"id":366002,"date":"2026-04-08T12:00:21","date_gmt":"2026-04-08T19:00:21","guid":{"rendered":"https:\/\/cms-articles.softonic.io\/es\/?p=415655"},"modified":"2026-04-08T12:00:48","modified_gmt":"2026-04-08T19:00:48","slug":"anthropics-new-ai-is-so-powerful-that-it-has-scared-its-ceo-they-will-not-launch-it","status":"publish","type":"post","link":"https:\/\/cms-articles.softonic.io\/en\/anthropics-new-ai-is-so-powerful-that-it-has-scared-its-ceo-they-will-not-launch-it\/","title":{"rendered":"Anthropic&#039;s new AI is so powerful that it has scared its CEO: they will not launch it"},"content":{"rendered":"\n<p class=\"wp-block-paragraph\">Anthropic has revealed its artificial intelligence model, Claude Mythos Preview, at a public event that has left both excitement and concern among experts.<\/p>\n\n\n<p class=\"wp-block-paragraph\">However, access to this powerful system will be limited exclusively to selected technology partners, which has generated controversy over its availability to the general public. And indeed, Claude is a very interesting AI.<\/p>\n\n\n<figure class=\"wp-block-embed is-type-rich is-provider-twitter wp-block-embed-twitter\"><div class=\"wp-block-embed__wrapper\">\n<blockquote class=\"twitter-tweet\" data-width=\"550\" data-dnt=\"true\"><p lang=\"en\" dir=\"ltr\">This is big&#8230; Anthropic just announced a model so powerful they won&#39;t release it to the public out of fear over the damage it will cause \ud83d\ude28<br><br>Claude Mythos Preview found thousands of zero-day exploits in every major operating system and web browser&#8230;<br><br>The numbers are hard to\u2026 <a href=\"https:\/\/t.co\/pEuokoHMA1\">https:\/\/t.co\/pEuokoHMA1<\/a> <a href=\"https:\/\/t.co\/FlQgGiavsd\">pic.twitter.com\/FlQgGiavsd<\/a><\/p>&mdash; Josh Kale (@JoshKale) <a href=\"https:\/\/twitter.com\/JoshKale\/status\/2041589742303649802?ref_src=twsrc%5Etfw\">April 7, 2026<\/a><\/blockquote><script async src=\"https:\/\/platform.twitter.com\/widgets.js\" charset=\"utf-8\"><\/script>\n<\/div><\/figure>\n\n\n<h2 class=\"wp-block-heading\">An AI that can be very dangerous right now<\/h2>\n\n\n<p class=\"wp-block-paragraph\"><strong>Anthropic&#8217;s decision is based on the model&#8217;s potential, which poses serious cybersecurity risks<\/strong>, capable of detecting and exploiting vulnerabilities in existing systems, as evidenced by a 27-year flaw in OpenBSD, a well-known high-security operating system.<\/p>\n\n\n<p class=\"wp-block-paragraph\">In terms of performance, Claude Mythos has set new standards by surpassing previous models like GPT 5.4 and Gemini 3.1 Pro in various benchmarks. In fact, in specific areas such as solving mathematical problems, Mythos approaches perfection.<\/p>\n\n\n<p class=\"wp-block-paragraph\">Its ability to provide coherent answers with a drastically lower hallucination rate than previous models also stands out, as <strong>this model can exhibit the necessary honesty by stating &#8220;I don&#8217;t know&#8221; when it lacks adequate information.<\/strong><\/p>\n\n\n<figure class=\"wp-block-embed is-type-rich is-provider-twitter wp-block-embed-twitter\"><div class=\"wp-block-embed__wrapper\">\n<blockquote class=\"twitter-tweet\" data-width=\"550\" data-dnt=\"true\"><p lang=\"en\" dir=\"ltr\">Anthropic says Project Glasswing uses its new Claude Mythos Preview model and has found security problems \u201cin every major operating system and web browser.\u201d <br><br>The model can scan code, test software, and identify serious vulnerabilities that may have been missed for years. <br><br>In\u2026 <a href=\"https:\/\/t.co\/6qQnQLoTdf\">pic.twitter.com\/6qQnQLoTdf<\/a><\/p>&mdash; Pirat_Nation \ud83d\udd34 (@Pirat_Nation) <a href=\"https:\/\/twitter.com\/Pirat_Nation\/status\/2041637894117585049?ref_src=twsrc%5Etfw\">April 7, 2026<\/a><\/blockquote><script async src=\"https:\/\/platform.twitter.com\/widgets.js\" charset=\"utf-8\"><\/script>\n<\/div><\/figure>\n\n\n<p class=\"wp-block-paragraph\">However, experts have expressed concern about a new phenomenon identified in the functioning of Mythos: although its errors are subtle and well-argued, they can lead to misunderstandings, causing even specialists to conduct thorough checks before accepting the model&#8217;s responses.<\/p>\n\n\n<p class=\"wp-block-paragraph\"><strong>To mitigate these risks, access to Claude Mythos will be restricted to a program known as the Glasswing Project,<\/strong> which will only include renowned technology partners such as AWS, Apple, Google, and Microsoft, who will use their capabilities to identify and correct potential vulnerabilities.<\/p>\n\n\n<p class=\"wp-block-paragraph\">Despite the impressive presentation of Mythos, Anthropic has indicated that this model represents only the beginning of a new era in the cybersecurity capabilities of language models. The company anticipates that there will be significant advancements in this field in the coming months and years, leaving open the possibility of even more sophisticated developments.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>Anthropic has revealed its artificial intelligence model, Claude Mythos Preview, at a public event that has generated both excitement and concern among experts. However, access to this powerful system will be limited exclusively to selected technology partners, which has sparked controversy over its availability to the general public. Claude is indeed a very interesting AI. An AI that right now can be very dangerous. Anthropic&#8217;s decision is based on the model&#8217;s potential, which poses serious cybersecurity risks, capable of detecting and exploiting vulnerabilities in existing systems, as evidenced by a [&hellip;]<\/p>\n","protected":false},"author":9317,"featured_media":366004,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":"","wpcf-pageviews":0},"categories":[1015],"tags":[],"usertag":[],"vertical":[],"content-category":[6771],"class_list":["post-366002","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-news","content-category-ai"],"aioseo_notices":[],"_links":{"self":[{"href":"https:\/\/cms-articles.softonic.io\/en\/wp-json\/wp\/v2\/posts\/366002","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/cms-articles.softonic.io\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/cms-articles.softonic.io\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/cms-articles.softonic.io\/en\/wp-json\/wp\/v2\/users\/9317"}],"replies":[{"embeddable":true,"href":"https:\/\/cms-articles.softonic.io\/en\/wp-json\/wp\/v2\/comments?post=366002"}],"version-history":[{"count":2,"href":"https:\/\/cms-articles.softonic.io\/en\/wp-json\/wp\/v2\/posts\/366002\/revisions"}],"predecessor-version":[{"id":366009,"href":"https:\/\/cms-articles.softonic.io\/en\/wp-json\/wp\/v2\/posts\/366002\/revisions\/366009"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/cms-articles.softonic.io\/en\/wp-json\/wp\/v2\/media\/366004"}],"wp:attachment":[{"href":"https:\/\/cms-articles.softonic.io\/en\/wp-json\/wp\/v2\/media?parent=366002"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/cms-articles.softonic.io\/en\/wp-json\/wp\/v2\/categories?post=366002"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/cms-articles.softonic.io\/en\/wp-json\/wp\/v2\/tags?post=366002"},{"taxonomy":"usertag","embeddable":true,"href":"https:\/\/cms-articles.softonic.io\/en\/wp-json\/wp\/v2\/usertag?post=366002"},{"taxonomy":"vertical","embeddable":true,"href":"https:\/\/cms-articles.softonic.io\/en\/wp-json\/wp\/v2\/vertical?post=366002"},{"taxonomy":"content-category","embeddable":true,"href":"https:\/\/cms-articles.softonic.io\/en\/wp-json\/wp\/v2\/content-category?post=366002"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}