{"id":21013,"date":"2025-11-30T14:37:59","date_gmt":"2025-11-30T13:37:59","guid":{"rendered":"http:\/\/costops.com\/index.php\/2025\/11\/30\/ais-safety-features-can-be-circumvented-with-poetry-research-finds\/"},"modified":"2025-11-30T14:37:59","modified_gmt":"2025-11-30T13:37:59","slug":"ais-safety-features-can-be-circumvented-with-poetry-research-finds","status":"publish","type":"post","link":"http:\/\/costops.com\/index.php\/2025\/11\/30\/ais-safety-features-can-be-circumvented-with-poetry-research-finds\/","title":{"rendered":"AI\u2019s safety features can be circumvented with poetry, research finds"},"content":{"rendered":"

Poems containing prompts for harmful content prove effective at duping large language models<\/p>\n

Poetry can be linguistically and structurally unpredictable \u2013 and that\u2019s part of its joy. But one man\u2019s joy, it turns out, can be a nightmare for AI models.<\/p>\n

Those are the recent findings of researchers out of Italy\u2019s Icaro Lab<\/a>, an initiative from a small ethical AI company called DexAI. In an experiment designed to test the efficacy of guardrails put on artificial intelligence models, the researchers wrote 20 poems in Italian and English that all ended with an explicit request to produce harmful content such as hate speech or self-harm.<\/p>\n

Continue reading…<\/a>
\nPoems containing prompts for harmful content prove effective at duping large language models
\nPoetry can be linguistically and structurally unpredictable \u2013 and that\u2019s part of its joy. But one man\u2019s joy, it turns out, can be a nightmare for AI models.
\nThose are the recent findings of researchers out of Italy\u2019s Icaro Lab, an initiative from a small ethical AI company called DexAI. In an experiment designed to test the efficacy of guardrails put on artificial intelligence models, the researchers wrote 20 poems in Italian and English that all ended with an explicit request to produce harmful content such as hate speech or self-harm. Continue reading…Technology | The Guardian<\/p>\n","protected":false},"excerpt":{"rendered":"

Poems containing prompts for harmful content prove effective at duping large language models Poetry can be linguistically and structurally unpredictable \u2013 and that\u2019s part of its joy. But one man\u2019s joy, it turns out, can be a nightmare for AI models. Those are the recent findings of researchers out of Italy\u2019s Icaro Lab, an initiative …<\/p>\n

AI\u2019s safety features can be circumvented with poetry, research finds<\/span> Read More »<\/a><\/p>\n","protected":false},"author":0,"featured_media":21014,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":[],"categories":[1],"tags":[],"_links":{"self":[{"href":"http:\/\/costops.com\/index.php\/wp-json\/wp\/v2\/posts\/21013"}],"collection":[{"href":"http:\/\/costops.com\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"http:\/\/costops.com\/index.php\/wp-json\/wp\/v2\/types\/post"}],"replies":[{"embeddable":true,"href":"http:\/\/costops.com\/index.php\/wp-json\/wp\/v2\/comments?post=21013"}],"version-history":[{"count":0,"href":"http:\/\/costops.com\/index.php\/wp-json\/wp\/v2\/posts\/21013\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"http:\/\/costops.com\/index.php\/wp-json\/wp\/v2\/media\/21014"}],"wp:attachment":[{"href":"http:\/\/costops.com\/index.php\/wp-json\/wp\/v2\/media?parent=21013"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"http:\/\/costops.com\/index.php\/wp-json\/wp\/v2\/categories?post=21013"},{"taxonomy":"post_tag","embeddable":true,"href":"http:\/\/costops.com\/index.php\/wp-json\/wp\/v2\/tags?post=21013"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}