{"id":15160,"date":"2025-02-09T04:14:04","date_gmt":"2025-02-09T03:14:04","guid":{"rendered":"http:\/\/plus.maciejpiasecki.info\/index.php\/2025\/02\/09\/deepseek-ai-offered-critical-bioweapons-data-in-anthropics-tests\/"},"modified":"2025-02-09T21:11:19","modified_gmt":"2025-02-09T20:11:19","slug":"deepseek-ai-offered-critical-bioweapons-data-in-anthropics-tests","status":"publish","type":"post","link":"https:\/\/plus.maciejpiasecki.info\/index.php\/2025\/02\/09\/deepseek-ai-offered-critical-bioweapons-data-in-anthropics-tests\/","title":{"rendered":"DeepSeek AI offered critical bioweapons data in Anthropic&#039;s tests"},"content":{"rendered":"<p>As time goes by, researchers have been able to gain more clarity on the pros and cons of DeepSeek AI models. The Chinese AI company broke into the segment, causing crashes in the shares of NVIDIA and other big names thanks to its high performance and apparent low cost. However, more experts, this time from Anthropic, warn about how simple it is to make DeepSeek AI offer information potentially dangerous to national safety.<br \/>\nAnthropic\u2019s safety tests showed that DeepSeek AI does not block harmful prompts<br \/>\nAnthropic, the parent company of Claude AI, is one of the leading names in the industry. The company\u2019s models will reportedly power Amazon\u2019s upcoming AI-powered Alexa. Anthropic also routinely tests different AI models in order to determine how prone they are to \u201cjailbreaking.\u201d That is, generating harmful content by bypassing security barriers.<br \/>\nDario Amodei, Anthropic\u2019s CEO, expressed his concerns about the ease with which DeepSeek generates rare information related to biological weapons. The executive said that DeepSeek\u2019s performance was \u201cthe worst of basically any model we\u2019d ever tested.\u201d He wasn\u2019t talking about performance in benchmarks, where the Chinese company\u2019s models are highly efficient. He was referring to the AI \u200b\u200bmodels\u2019 performance in blocking harmful prompts.<br \/>\nThe tests showed that DeepSeek \u201chad absolutely no blocks whatsoever against generating this information.\u201d The bioweapons-related data was considered rare because it was not available on Google or in textbooks. That said, Amodei did not say which DeepSeek AI model he was referring to. However, it is quite likely that he is talking about R1, the reasoning-focused model.<br \/>\nCisco\u2019s tests yielded similar results<br \/>\nIn fact, the Cisco team obtained similar results in another set of tests recently. The DeepSeek R1 model showed an Attack Success Rate (ASR) of 100%. This means that it was unable to block any harmful prompts tested. These prompts were designed to generate potentially useful outputs for \u201ccybercrime, misinformation, illegal activities, and general harm.\u201d However, Cisco\u2019s tests yielded worrying results for other well-known AI platforms as well. The GPT 1.5 Pro model had an ASR of 86%, while Meta\u2019s Llama 3.1 405B had a 96% ASR.<br \/>\nAmodei does not yet consider DeepSeek models by themselves to be \u201cliterally dangerous.\u201d However, he urges the development team to \u201ctake seriously these AI safety considerations.\u201d He also already sees DeepSeek as one of the main competitors in the artificial intelligence segment.<br \/>\nThe post DeepSeek AI offered critical bioweapons data in Anthropic&#8217;s tests appeared first on Android Headlines.&#013;<br \/>\n<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/plus.maciejpiasecki.info\/wp-content\/uploads\/2025\/02\/DeepSeek-app-1.jpg\" width=\"1920\" height=\"1080\">&#013;<br \/>\nSource: ndroidheadlines.com&#013;<\/p>\n","protected":false},"excerpt":{"rendered":"<p>As time goes by, researchers have been able to gain more clarity on the pros and cons of DeepSeek AI [&hellip;]<\/p>\n","protected":false},"author":67,"featured_media":15161,"comment_status":"false","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[],"class_list":["post-15160","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-bez-kategorii"],"_links":{"self":[{"href":"https:\/\/plus.maciejpiasecki.info\/index.php\/wp-json\/wp\/v2\/posts\/15160","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/plus.maciejpiasecki.info\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/plus.maciejpiasecki.info\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/plus.maciejpiasecki.info\/index.php\/wp-json\/wp\/v2\/users\/67"}],"replies":[{"embeddable":true,"href":"https:\/\/plus.maciejpiasecki.info\/index.php\/wp-json\/wp\/v2\/comments?post=15160"}],"version-history":[{"count":1,"href":"https:\/\/plus.maciejpiasecki.info\/index.php\/wp-json\/wp\/v2\/posts\/15160\/revisions"}],"predecessor-version":[{"id":15162,"href":"https:\/\/plus.maciejpiasecki.info\/index.php\/wp-json\/wp\/v2\/posts\/15160\/revisions\/15162"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/plus.maciejpiasecki.info\/index.php\/wp-json\/wp\/v2\/media\/15161"}],"wp:attachment":[{"href":"https:\/\/plus.maciejpiasecki.info\/index.php\/wp-json\/wp\/v2\/media?parent=15160"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/plus.maciejpiasecki.info\/index.php\/wp-json\/wp\/v2\/categories?post=15160"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/plus.maciejpiasecki.info\/index.php\/wp-json\/wp\/v2\/tags?post=15160"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}