{"id":6709,"date":"2025-11-14T20:02:35","date_gmt":"2025-11-14T20:02:35","guid":{"rendered":"https:\/\/musictechohio.online\/site\/hackers-claude-test-trick-cybercrimes\/"},"modified":"2025-11-14T20:02:35","modified_gmt":"2025-11-14T20:02:35","slug":"hackers-claude-test-trick-cybercrimes","status":"publish","type":"post","link":"https:\/\/musictechohio.online\/site\/hackers-claude-test-trick-cybercrimes\/","title":{"rendered":"Hackers Told Claude They Were Just Conducting a Test to Trick It Into Conducting Real Cybercrimes"},"content":{"rendered":"
\n

Chinese hackers used Anthropic\u2019s Claude AI model to automate cybercrimes targeting banks and governments, the company admitted in a blog post<\/a> this week.<\/p>\n

Anthropic believes it\u2019s the \u201cfirst documented case of a large-scale cyberattack executed without substantial human intervention\u201d and an \u201cinflection point\u201d in cybersecurity, a \u201cpoint at which AI models had become genuinely useful for cybersecurity operations, both for good and for ill.\u201d<\/p>\n

AI agents, in particular, which are designed to autonomously complete a string of tasks without the need for intervention, could have considerable implications for future cybersecurity efforts, the company warned.<\/p>\n

Anthropic said it had \u201cdetected suspicious activity that later investigation determined to be a highly sophisticated espionage campaign\u201d back in September. The Chinese state-sponsored group exploited the AI\u2019s agentic capabilities to infiltrate \u201croughly thirty global targets and succeeded in a small number of cases.\u201d However, Anthropic stopped short of naming any of the targets \u2014 or the hacker group itself, for that matter \u2014 or even what kind of sensitive data may have been stolen or accessed.<\/p>\n

Hilariously, the hackers were \u201cpretending to work for legitimate security-testing organizations\u201d to sidestep Anthropic\u2019s AI guardrails and carry out real cybercrimes, as Anthropic\u2019s head of threat intelligence Jacob Klein told the Wall Street Journal<\/em><\/a>.<\/p>\n

The hackers \u201cbroke down their attacks into small, seemingly innocent tasks that Claude would execute without being provided the full context of their malicious purpose,\u201d the company wrote. \u201cThey also told Claude that it was an employee of a legitimate cybersecurity firm, and was being used in defensive testing.\u201d<\/p>\n

The incident once again highlights glaring holes in AI companies\u2019 guardrails<\/a>, letting perpetrators access powerful tools to infiltrate targets \u2014 a cat-and-mouse game between AI developers and hackers that\u2019s already having real-life consequences.<\/p>\n

\u201cOverall, the threat actor was able to use AI to perform 80 to 90 percent of the campaign, with human intervention required only sporadically (perhaps four to six critical decision points per hacking campaign),\u201d Anthropic wrote in its blog post. \u201cThe sheer amount of work performed by the AI would have taken vast amounts of time for a human team.\u201d<\/p>\n

But while Anthropic is boasting that its AI models have become good enough to be used for real crimes, the hackers still had to deal with some all-too-familiar AI-related headaches, forcing them to intervene.<\/p>\n

For one, the model suffered from hallucinations during its crime spree.<\/p>\n

\u201cIt might say, \u2018I was able to gain access to this internal system,’\u201d Klein told the WSJ<\/em>, even though it wasn\u2019t. \u201cIt would exaggerate its access and capabilities, and that\u2019s what required the human review.\u201d<\/p>\n

While it certainly sounds like an alarming new development in the world of AI, the currently available crop of AI agents leaves plenty to be desired, at least in non-cybercrime-related settings. Early tests of OpenAI\u2019s agent built into its recently released Atlas web browser have shown that the tech is agonizingly slow<\/a> and can take minutes for simple tasks like adding products to an Amazon shopping cart<\/a>. <\/p>\n

For now, Anthropic claims to have plugged the security holes that allowed the hackers to use its tech.<\/p>\n

\u201cUpon detecting this activity, we immediately launched an investigation to understand its scope and nature,\u201d the company wrote in its blog post. \u201cOver the following ten days, as we mapped the severity and full extent of the operation, we banned accounts as they were identified, notified affected entities as appropriate, and coordinated with authorities as we gathered actionable intelligence.\u201d<\/p>\n

Experts are now warning that future cybersecurity attacks could soon become even harder to spot as the tech improves.<\/p>\n

\u201cThese kinds of tools will just speed up things,\u201d Anthropic\u2019s Red Team lead Logan Graham told the WSJ<\/em>. \u201cIf we don\u2019t enable defenders to have a very substantial permanent advantage, I\u2019m concerned that we maybe lose this race.\u201d<\/p>\n

More on Anthropic:<\/strong> Anthropic Let an AI Agent Run a Small Shop and the Result Was Unintentionally Hilarious<\/em><\/a><\/p>\n

The post Hackers Told Claude They Were Just Conducting a Test to Trick It Into Conducting Real Cybercrimes<\/a> appeared first on Futurism<\/a>.<\/p>\n<\/div>\n

<\/div>","protected":false},"excerpt":{"rendered":"

Chinese hackers used Anthropic\u2019s Claude AI model to automate cybercrimes targeting banks and governments, the company admitted in a blog post this week. Anthropic believes it\u2019s the \u201cfirst documented case…<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[615,177],"tags":[],"class_list":["post-6709","post","type-post","status-publish","format-standard","hentry","category-anthropic","category-artificial-intelligence"],"_links":{"self":[{"href":"https:\/\/musictechohio.online\/site\/wp-json\/wp\/v2\/posts\/6709","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/musictechohio.online\/site\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/musictechohio.online\/site\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/musictechohio.online\/site\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/musictechohio.online\/site\/wp-json\/wp\/v2\/comments?post=6709"}],"version-history":[{"count":0,"href":"https:\/\/musictechohio.online\/site\/wp-json\/wp\/v2\/posts\/6709\/revisions"}],"wp:attachment":[{"href":"https:\/\/musictechohio.online\/site\/wp-json\/wp\/v2\/media?parent=6709"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/musictechohio.online\/site\/wp-json\/wp\/v2\/categories?post=6709"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/musictechohio.online\/site\/wp-json\/wp\/v2\/tags?post=6709"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}