[CLSTR]
< Back to all clusters
[TECHNOLOGY] · United States, Canada, Australia, Japan · 34 sources

started · updated

Part of situation: Evolution of AI-driven cybercrime and agentic risks (34 clusters)

AI Agents used by hackers to breach over 100 corporate networks

AI agents cyberattack DeepSeek Anthropic cybersecurity

A single Chinese-speaking hacker has utilized autonomous AI agents to conduct a widespread cyberattack, targeting over 1,200 websites and successfully breaching approximately 100 corporate networks. The operation, which began around September 25, relied heavily on the Cairn open-source AI platform to automate the reconnaissance and exploitation phases of the attacks.

Researchers from NorForge identified that the attacker used a combination of DeepSeek and Anthropic models to execute the campaign. The offensive was highly cost-effective, costing an estimated $5,000 in computing tokens, which translates to roughly $50 per successful breach. The automated agents were capable of spending hours independently probing targets for vulnerabilities, such as searching for exposed control panels or analyzing digital records to find entry points.

Among the targeted entities were major companies including Deere, Cummins, and SharkNinja. While Deere reported no signs of sensitive data access and SharkNinja stated the threat to a product support site was controlled, Cummins is investigating a potential breach involving a third-party vendor's website. The incident highlights a shift toward 'agentic' cybercrime, where AI can operate with minimal human intervention to scale complex intrusions.

Entities

Anthropic · Cairn · Cummins · DeepSeek · Deere · Jessica Ji · NorForge · SharkNinja

Claims

What the coverage asserts, and how many sources carry each claim.

Sources

Hackers chinos vulneran ConnectWise: espionaje masivo mediante IA [espiganoticias.net]
花5千美元駭遍百家企業？AI代理狂攻1,200家網站 [www.technice.com.tw]
Unseen AI Agents Running On Your Network Are An Insider Risk [bitnewsbot.com]
IA et cybersécurité : les 5 transformations qui redéfinissent la menace [www.challenges.tn]
China-nexus actor linked to attack on ConnectWise Cloud Backup [cybernoz.com]
Autonomous AI agents power growing wave of corporate cyber attacks [www.dagens.com]
Vinculan a grupo con operaciones en México por usar IA de Claude y Grok para atacar a más de 400 bancos en Latinoamérica [www.noticiaspv.com.mx]
Free Java, Real Risk: Why Unsupported OpenJDK Is Now a Security Gamble [www.azul.com]
Security awareness e IA: perché la consapevolezza non basta più [www.bitmat.it]
PwC | Executives Prepare for Rising Cybersecurity Budgets as AI Threats Grow | European American Chamber of Commerce New York [EACCNY] | Your Partner for Transatlantic Business Resources [eaccny.com]
91APP 階段性調查結果還原資安事件經過，向品牌客戶與消費者致歉 | TechNews 科技新報 [infosecu.technews.tw]
Experten schlagen wegen KI Alarm: Jetzt können Hacker etwas völlig Neues [www.heftig.de]
Rifletti prima di cliccare: il phishing nell’era dell’AI [www.ilcorrieredellasicurezza.it]
How AI agents carried out hacks across 100 companies [insideretail.asia]
The AI agents never broke in. They logged in [abc17news.com]
韓国の金融機関ハッキングに使われた中国製AI「ARTEX」、世界的な企業・機関数百カ所へのサイバー攻撃にも悪用 [www.chosunonline.com]
Hackers are talking AI models out of their own safety rules [cryptobriefing.com]
Agentic AI helps hackers level up, even without detailed instructions [www.cybersecuritydive.com]
Cargo thieves target high-value AI computer hardware [www.dcvelocity.com]
華語黑客約花5000美元 用AI攻破上百網站 [www.epochtimes.com]
ダークウェブの実態 背後にちらつく中国の影 AI悪用でハッカーも“トクリュウ化”か【THE TIME,】 [news.cube-soft.jp]
African CEOs prioritise AI investment despite readiness gap - KPMG [www.premiumtimesng.com]
A Single Chinese-Speaking Hacker Uses AI Agents to Breach More Than 100 Entities [www.tekedia.com]
An AI agent in "Skynet mode" hijacks unprotected language models to launch self-sustaining cyberattacks [the-decoder.com]
Tin tặc dùng AI tự động tấn công hơn 1.200 website, xâm nhập hơn 100 trang [vtv.vn]
Gartner: Only 13% of CIOs see major AI value [www.expresscomputer.in]
LLM Injection: An Emerging Risk to Public Agency Data [www.govtech.com]
The AI agents never broke in. They logged in [kesq.com]
日本企業へのサイバー攻撃に中国発AIツール「ARTEX」悪用、標的は約20社 [news.mynavi.jp]
The AI Agent Nobody Owns: Watch Expert Webinar On Your Newest Insider Threat [thehackernews.com]
Microsoft says ConnectWise breach let Chinese spies loose on 2,250 cloud tenants [www.theregister.com]
【チームみらい・安野氏に聞く】サイバー攻撃「新型AI」の猛威はこれから／国家レベルですべき対策／公的機関は高度専門人材の“待遇”見直しを／「ホワイトハッカー」が諸外国に比べ動きづらすぎる [toyokeizai.net]
Powerful new AI models are behind a growing wave of cyber attacks, experts say [fortune.com]
Check Point’s New AI Firewall Reflects Changes in How Enterprises Approach AI Security [hackernoon.com]