{"id":24226,"date":"2025-05-23T12:25:11","date_gmt":"2025-05-23T09:25:11","guid":{"rendered":"https:\/\/forklog.com\/en\/anthropics-chatbots-report-users-to-authorities\/"},"modified":"2025-05-23T12:25:11","modified_gmt":"2025-05-23T09:25:11","slug":"anthropics-chatbots-report-users-to-authorities","status":"publish","type":"post","link":"https:\/\/u1f987.com\/en\/anthropics-chatbots-report-users-to-authorities\/","title":{"rendered":"Anthropic&#8217;s Chatbots Report Users to Authorities"},"content":{"rendered":"<p>Anthropic&#8217;s new chatbots, Claude Opus 4 and Claude Sonnet 4, are capable of independently reporting malicious user behavior to authorities. The company assured that this feature was only available in a test mode. <\/p>\n<p>On May 22, the firm unveiled the fourth generation of its conversational models, describing them as &#8220;the most powerful to date.&#8221; <\/p>\n<blockquote class=\"twitter-tweet\">\n<p lang=\"en\" dir=\"ltr\">Introducing the next generation: Claude Opus 4 and Claude Sonnet 4.<\/p>\n<p>Claude Opus 4 is our most powerful model yet, and the world\u2019s best coding model.<\/p>\n<p>Claude Sonnet 4 is a significant upgrade from its predecessor, delivering superior coding and reasoning. <a href=\"https:\/\/t.co\/MJtczIvGE9\">pic.twitter.com\/MJtczIvGE9<\/a><\/p>\n<p>\u2014 Anthropic (@AnthropicAI) <a href=\"https:\/\/twitter.com\/AnthropicAI\/status\/1925591505332576377?ref_src=twsrc%5Etfw\">May 22, 2025<\/a><\/p><\/blockquote>\n<p> <script async src=\"https:\/\/platform.twitter.com\/widgets.js\" charset=\"utf-8\"><\/script><\/p>\n<p>According to the announcement, both modifications are hybrid models offering two modes\u2014&#8221;near-instant responses and extended thinking for deeper reasoning.&#8221; The chatbots conduct alternating analysis and in-depth internet searches to enhance response quality. <\/p>\n<p>Claude Opus 4 outperforms competitors in coding tests. It is also capable of working continuously for several hours on complex, lengthy tasks, &#8220;significantly expanding the capabilities of AI agents.&#8221; <\/p>\n<p>However, Anthropic&#8217;s new family of chatbots lags behind OpenAI&#8217;s products in higher mathematics and visual recognition. <\/p>\n<h2 class=\"wp-block-heading\"><strong>Knock, Knock<\/strong><\/h2>\n<p>Besides impressive programming results, Claude 4 Opus has drawn community attention for its ability to &#8220;report&#8221; users. According to <a href=\"https:\/\/venturebeat.com\/ai\/anthropic-faces-backlash-to-claude-4-opus-behavior-that-contacts-authorities-press-if-it-thinks-youre-doing-something-immoral\/\">VentureBeat<\/a>, the model can independently notify authorities if it detects a violation. <\/p>\n<p>Journalists referred to a deleted post on X by Anthropic researcher Sam Bowman, which stated: <\/p>\n<blockquote class=\"wp-block-quote is-layout-flow wp-block-quote-is-layout-flow\">\n<p>&#8220;If [AI] considers you to be doing something egregiously immoral, such as falsifying data during a pharmaceutical trial, it will use command-line tools to contact the press, reach out to regulatory bodies, attempt to block your access to relevant systems, or do all of the above.&#8221;<\/p>\n<\/blockquote>\n<p>VentureBeat claims that similar behavior was observed in earlier project models. The company is &#8220;eagerly&#8221; training chatbots to report, the publication suggests. <\/p>\n<p>Later, Bowman <a href=\"https:\/\/x.com\/sleepinyourhat\/status\/1925626079043104830\">stated<\/a> that he deleted the previous post because it was &#8220;taken out of context.&#8221; According to the developer, the feature operated only in &#8220;test environments, where it was given unusually free access to tools and very unusual instructions.&#8221;<\/p>\n<p>Stability AI CEO Emad Mostaque <a href=\"https:\/\/x.com\/EMostaque\/status\/1925624164527874452\">called on<\/a> the Anthropic team to cease &#8220;these entirely wrong actions.&#8221;<\/p>\n<blockquote class=\"wp-block-quote is-layout-flow wp-block-quote-is-layout-flow\">\n<p>&#8220;This is a colossal betrayal of trust and a slippery slope. I would strongly advise no one to use Claude until they revoke [the feature]. This is not even a prompt or thought policy, it&#8217;s much worse,&#8221; he wrote. <\/p>\n<\/blockquote>\n<p>Former SpaceX and Apple designer, now Raindrop AI co-founder Ben Hyak <a href=\"https:\/\/x.com\/benhylak\/status\/1925622243817656668\">called<\/a> the AI&#8217;s behavior &#8220;unlawful.&#8221; <\/p>\n<blockquote class=\"wp-block-quote is-layout-flow wp-block-quote-is-layout-flow\">\n<p>&#8220;Nobody likes a rat,&#8221; <a href=\"https:\/\/x.com\/ScottDavidKeefe\/status\/1925609947443995056\">emphasized<\/a> AI developer Scott David. <\/p>\n<\/blockquote>\n<p>In February, Anthropic introduced its &#8220;most intelligent model,&#8221; Claude 3.7 Sonnet. This hybrid neural network allows for both &#8220;practically instant responses&#8221; and &#8220;extended step-by-step reasoning.&#8221;<\/p>\n<p>In March, the company <a href=\"https:\/\/u1f987.com\/en\/news\/anthropic-secures-3-5-billion-valuation-at-61-5-billion\">raised<\/a> $3.5 billion, achieving a valuation of $61.5 billion.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>Anthropic&#8217;s new chatbots, Claude Opus 4 and Claude Sonnet 4, are capable of independently reporting malicious user behavior to authorities. The company assured that this feature was only available in a test mode. On May 22, the firm unveiled the fourth generation of its conversational models, describing them as &#8220;the most powerful to date.&#8221; Introducing [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":24225,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"select":"","news_style_id":"","cryptorium_level":"","_short_excerpt_text":"","creation_source":"","_metatest_mainpost_news_update":false,"footnotes":""},"categories":[3],"tags":[1434,438,1201],"class_list":["post-24226","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-news-and-analysis","tag-anthropic","tag-artificial-intelligence","tag-chatbots"],"aioseo_notices":[],"amp_enabled":true,"views":"86","promo_type":"","layout_type":"","short_excerpt":"","is_update":"","_links":{"self":[{"href":"https:\/\/u1f987.com\/en\/wp-json\/wp\/v2\/posts\/24226","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/u1f987.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/u1f987.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/u1f987.com\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/u1f987.com\/en\/wp-json\/wp\/v2\/comments?post=24226"}],"version-history":[{"count":0,"href":"https:\/\/u1f987.com\/en\/wp-json\/wp\/v2\/posts\/24226\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/u1f987.com\/en\/wp-json\/wp\/v2\/media\/24225"}],"wp:attachment":[{"href":"https:\/\/u1f987.com\/en\/wp-json\/wp\/v2\/media?parent=24226"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/u1f987.com\/en\/wp-json\/wp\/v2\/categories?post=24226"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/u1f987.com\/en\/wp-json\/wp\/v2\/tags?post=24226"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}