[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"article-trm-labs-ai-agents-crypto-investigations-zh":3,"tags-trm-labs-ai-agents-crypto-investigations-zh":31,"related-lang-trm-labs-ai-agents-crypto-investigations-zh":42,"related-posts-trm-labs-ai-agents-crypto-investigations-zh":46,"series-blockchain-7ff10146-4ca0-4670-a02c-384dde04f610":83},{"id":4,"title":5,"content":6,"summary":7,"source":8,"source_url":9,"author":10,"image_url":11,"keywords":12,"language":19,"translated_content":10,"views":20,"is_premium":21,"created_at":22,"updated_at":22,"cover_image":11,"published_at":23,"rewrite_status":24,"rewrite_error":10,"rewritten_from_id":25,"slug":26,"category":27,"related_article_id":28,"status":29,"google_indexed_at":30,"x_posted_at":10,"tweet_text":10,"title_rewritten_at":10,"title_original":10,"key_takeaways":10,"topic_cluster_id":10,"embedding":10,"is_canonical_seed":21},"7ff10146-4ca0-4670-a02c-384dde04f610","TRM Labs 將 AI agent 帶進加密調查","\u003Cp>\u003Ca href=\"https:\u002F\u002Fwww.trmlabs.com\" target=\"_blank\" rel=\"noopener\">TRM Labs\u003C\u002Fa> 這次不是做一個聊天機器人而已。它是把 \u003Ca href=\"\u002Fnews\u002Ftrust-wallet-ai-agents-crypto-trades-zh\">AI\u003C\u002Fa> agent 直接塞進 \u003Ca href=\"https:\u002F\u002Fwww.trmlabs.com\u002Fproducts\u002Ftrm-forensics\" target=\"_blank\" rel=\"noopener\">TRM Forensics\u003C\u002Fa>。講白了，就是讓調查員用白話問問題，系統自己去跑區塊鏈追蹤。\u003C\u002Fp>\u003Cp>TRM 說，去年非法加密貨幣交易量到 \u003Cstrong>1580 億美元\u003C\u002Fstrong>。同時，\u003Ca href=\"\u002Fnews\u002Fai-agents-moving-into-real-work-zh\">AI\u003C\u002Fa> 相關詐騙和騙局也暴增 \u003Cstrong>500%\u003C\u002Fstrong>。這兩個數字放一起看，很直白：犯罪側已經在用自動化，防守方也只能跟上。\u003C\u002Fp>\u003Cp>這種工具不是給一般玩家玩的。它是給執法單位、金融機構，還有加密公司用的。重點不是炫技。重點是把原本很硬的分析流程，縮短成幾個自然語句。\u003C\u002Fp>\u003Ch2>TRM 這次到底上了什麼\u003C\u002Fh2>\u003Cp>TRM 的新 ag\u003Ca href=\"\u002Fnews\u002Frtk-cuts-claude-code-token-spend-zh\">en\u003C\u002Fa>t，核心功能很單純。你可以直接問它資金去了哪裡。也可以問它跨了哪些鏈。還可以問某個錢包群組，是否碰過已知的非法基礎設施。\u003C\u002Fp>\n\u003Cfigure class=\"my-6\">\u003Cimg src=\"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1775058149786-o0ok.png\" alt=\"TRM Labs 將 AI agent 帶進加密調查\" class=\"rounded-xl w-full\" loading=\"lazy\" \u002F>\u003C\u002Ffigure>\n\u003Cp>以前這種事，常常要靠資深分析師手寫查詢。那不是不能做。只是很花時間。尤其當案件一多，手邊又有多條鏈要查時，光是整理問題就會卡住。\u003C\u002Fp>\u003Cp>TRM 的做法，是把自然語言轉成調查動作。這表示調查員不用先學會複雜語法，再去查資料。對實務工作來說，這很省事。真的，這種設計比很多空泛的 AI 口號有用多了。\u003C\u002Fp>\u003Cul>\u003Cli>TRM 說非法加密交易量達 \u003Cstrong>1580 億美元\u003C\u002Fstrong>\u003C\u002Fli>\u003Cli>AI 詐騙與騙局增加 \u003Cstrong>500%\u003C\u002Fstrong>\u003C\u002Fli>\u003Cli>工具先給既有用戶使用\u003C\u002Fli>\u003Cli>目標族群是執法、金融與加密企業\u003C\u002Fli>\u003Cli>查詢方式改成白話提問\u003C\u002Fli>\u003C\u002Ful>\u003Cp>這種產品路線，也反映一個現實。區塊鏈分析工具很強，但學習成本高。你如果要先懂資料結構、節點關係、標記規則，才會問問題，那很多團隊根本用不滿。\u003C\u002Fp>\u003Cp>所以這次不是單純加一個 AI 功能。它是在改工作流。從「先學工具」變成「先問問題」。這差很多。\u003C\u002Fp>\u003Ch2>為什麼現在才上 AI agent\u003C\u002Fh2>\u003Cp>原因很簡單。案件量變多了，分析人力沒那麼快補上。TRM 的說法也很直接：工作量成長速度，比人力快。\u003C\u002Fp>\u003Cp>TRM 的 \u003Ca href=\"https:\u002F\u002Fwww.trmlabs.com\u002Fabout\" target=\"_blank\" rel=\"noopener\">Ari Redbord\u003C\u002Fa> 在說明裡講得很白：調查員要同時處理多條區塊鏈、多個司法管轄區，還有不同類型的犯罪手法。這不是單一工具能硬扛的。\u003C\u002Fp>\u003Cblockquote>“What we’re seeing every day is that the caseload is growing faster than the workforce, and investigators are being asked to operate across dozens of blockchains, jurisdictions, and typologies simultaneously.” — \u003Ca href=\"https:\u002F\u002Fwww.trmlabs.com\u002Fabout\" target=\"_blank\" rel=\"noopener\">Ari Redbord\u003C\u002Fa>\u003C\u002Fblockquote>\u003Cp>這句話很重要。因為它點出 AI 在這裡的定位。不是取代調查員。也不是自動判案。它是把重複查詢、初步追蹤、資料整理這些雜事先做掉。\u003C\u002Fp>\u003Cp>而且現在的犯罪流程也更工業化了。以前可能是一個人搞一個錢包。現在常常是 bot、假帳號、深偽語音、洗錢跳板一起上。你要是還用舊方法追，真的會追到手軟。\u003C\u002Fp>\u003Ch2>跟其他區塊鏈工具比，差在哪\u003C\u002Fh2>\u003Cp>這個市場不是只有 TRM 在玩。\u003Ca href=\"https:\u002F\u002Fwww.chainalysis.com\" target=\"_blank\" rel=\"noopener\">Chainalysis\u003C\u002Fa>、\u003Ca href=\"https:\u002F\u002Fwww.elliptic.co\" target=\"_blank\" rel=\"noopener\">Elliptic\u003C\u002Fa>、\u003Ca href=\"https:\u002F\u002Fwww.ciphertrace.com\" target=\"_blank\" rel=\"noopener\">CipherTrace\u003C\u002Fa> 都在做區塊鏈分析、風險標記、AML 追查這類工作。\u003C\u002Fp>\n\u003Cfigure class=\"my-6\">\u003Cimg src=\"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1775058168940-8msj.png\" alt=\"TRM Labs 將 AI agent 帶進加密調查\" class=\"rounded-xl w-full\" loading=\"lazy\" \u002F>\u003C\u002Ffigure>\n\u003Cp>差別在於，TRM 這次多了一層對話介面。也就是說，你不用先熟悉整套查詢邏輯，先用自然語言說需求就好。對新手來說，這很友善。對老手來說，這是省時間。\u003C\u002Fp>\u003Cp>我覺得這種差異很實際。因為很多分析軟體的問題，不是功能不夠，而是太難用。功能再強，沒人會問，最後還是放著長灰塵。\u003C\u002Fp>\u003Cul>\u003Cli>\u003Ca href=\"https:\u002F\u002Fwww.chainalysis.com\" target=\"_blank\" rel=\"noopener\">Chainalysis\u003C\u002Fa> 偏重監控、合規與調查\u003C\u002Fli>\u003Cli>\u003Ca href=\"https:\u002F\u002Fwww.elliptic.co\" target=\"_blank\" rel=\"noopener\">Elliptic\u003C\u002Fa> 強調風險評分與暴露分析\u003C\u002Fli>\u003Cli>\u003Ca href=\"https:\u002F\u002Fwww.ciphertrace.com\" target=\"_blank\" rel=\"noopener\">CipherTrace\u003C\u002Fa> 長期做 AML 與偵查流程\u003C\u002Fli>\u003Cli>\u003Ca href=\"https:\u002F\u002Fwww.trmlabs.com\u002Fproducts\u002Ftrm-forensics\" target=\"_blank\" rel=\"noopener\">TRM Forensics\u003C\u002Fa> 現在加入 prompt 式調查\u003C\u002Fli>\u003C\u002Ful>\u003Cp>如果把這幾家放一起看，趨勢很清楚。下一輪競爭，不只是誰追錢包最快。還包括誰能讓調查員最快問對問題，然後把答案整理成可用證據。\u003C\u002Fp>\u003Cp>這對公部門很重要。因為人力通常不夠。對交易所和支付業者也重要。因為他們要更快抓出可疑資金流，少漏掉一筆，就少一個大洞。\u003C\u002Fp>\u003Ch2>這對加密執法意味著什麼\u003C\u002Fh2>\u003Cp>先講現實版答案：AI agent 不會自己抓罪犯。它還是可能看錯脈絡，也可能漏掉邊界案例。更麻煩的是，它如果講得很像真的，反而容易誤導人。\u003C\u002Fp>\u003Cp>所以在執法場景裡，人類審核還是不能少。AI 可以幫你縮短前處理時間。它可以幫你整理鏈上路徑。它也可以幫你把調查起點拉出來。但最後下結論的人，還是得是人。\u003C\u002Fp>\u003Cp>不過方向已經很明顯了。調查工具正在變成分析助理，而不是純資料庫介面。這種改法，會讓查詢更快，也會讓分析師把時間放回判斷本身。\u003C\u002Fp>\u003Cp>如果 TRM 的做法跑得順，其他廠商大概也會跟進。到時候，prompt 式調查可能會變成標配。問題只剩一個：誰能把速度和可稽核性一起做好。\u003C\u002Fp>\u003Ch2>這波背後的產業脈絡\u003C\u002Fh2>\u003Cp>區塊鏈分析這個市場，這幾年一直在往兩個方向走。第一個是合規。第二個是執法。前者看風險，後者看證據。兩邊都很吃資料品質，也都很吃流程效率。\u003C\u002Fp>\u003Cp>AI agent 的加入，代表產品介面開始往自然語言靠攏。這跟一般企業軟體的趨勢一樣。大家都想少學幾個指令，多做一點判斷。說穿了，沒人想把時間浪費在記語法。\u003C\u002Fp>\u003Cp>但這裡有個雷點。越是方便的工具，越要保留紀錄。因為調查不是聊天。每一步都可能進報告，甚至進法庭。能不能重現、能不能解釋，會比回答得快不快更重要。\u003C\u002Fp>\u003Ch2>接下來我會看什麼\u003C\u002Fh2>\u003Cp>我會先看兩件事。第一，這個 agent 會不會真的減少分析時間。第二，它產出的結果，能不能維持可追蹤、可驗證、可審計。\u003C\u002Fp>\u003Cp>如果 TRM 做得好，下一步很可能是更多國家執法單位跟進。金融機構也會更想買。畢竟，當犯罪側已經會用 AI，防守側不會只靠人工慢慢查。\u003C\u002Fp>\u003Cp>你可能會想問，這會不會變成所有區塊鏈工具的標配？我覺得很有機會。只是最後活下來的，不會是最會講 AI 的廠商，而是最能把 AI 關進流程裡的那一家。\u003C\u002Fp>","TRM Labs 把 AI agent 放進 TRM Forensics，讓調查員用白話查區塊鏈金流。公司稱去年非法加密交易達 1580 億美元，AI 詐騙也暴增 500%。","www.coindesk.com","https:\u002F\u002Fwww.coindesk.com\u002Fpolicy\u002F2026\u002F03\u002F25\u002Fai-agents-to-help-investigators-unearth-crypto-criminals-according-to-new-trm-program",null,"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1775058149786-o0ok.png",[13,14,15,16,17,18],"TRM Labs","AI agent","區塊鏈分析","加密貨幣調查","TRM Forensics","crypto investigations","zh",1,false,"2026-04-01T10:33:30.166266+00:00","2026-04-01T10:33:30.127+00:00","done","56c77a57-cc01-4901-a5b8-aa4d06a116fe","trm-labs-ai-agents-crypto-investigations-zh","blockchain","468d5c05-99ac-4115-ae8e-842755509672","published","2026-04-09T09:00:53.597+00:00",[32,33,35,37,39,40],{"name":16,"slug":16},{"name":18,"slug":34},"crypto-investigations",{"name":13,"slug":36},"trm-labs",{"name":14,"slug":38},"ai-agent",{"name":15,"slug":15},{"name":17,"slug":41},"trm-forensics",{"id":28,"slug":43,"title":44,"language":45},"trm-labs-ai-agents-crypto-investigations-en","TRM Labs adds AI agents for crypto investigations","en",[47,53,59,65,71,77],{"id":48,"slug":49,"title":50,"cover_image":51,"image_url":51,"created_at":52,"category":27},"8c37fa14-a081-4810-b5b8-2a2a184a7d1d","web3-communication-trust-infrastructure-2026-zh","Web3 溝通正在變成信任基礎設施","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1778797251989-it0w.png","2026-05-14T22:20:32.600359+00:00",{"id":54,"slug":55,"title":56,"cover_image":57,"image_url":57,"created_at":58,"category":27},"9059e494-8f72-4c34-a888-2424c682da10","why-bases-x402-protocol-matters-more-than-100m-zh","為什麼 Base 的 x402 協議比 1 億美元里程碑更重要","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1778719260627-a0va.png","2026-05-14T00:40:19.962138+00:00",{"id":60,"slug":61,"title":62,"cover_image":63,"image_url":63,"created_at":64,"category":27},"74969a5b-7ec5-4686-80ee-fa39a5cc43d4","gala-games-web3-gaming-2026-zh","Gala Games 在 Web3 遊戲找回存在感","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1778689265110-p0y5.png","2026-05-13T16:20:41.782583+00:00",{"id":66,"slug":67,"title":68,"cover_image":69,"image_url":69,"created_at":70,"category":27},"d330d44a-4eff-4ba6-aa72-5ef246e31c64","why-lace-20-matters-more-than-cardanos-next-hard-fork-zh","為什麼 Lace 2.0 比 Cardano 下一次硬分叉更重要","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1778681462051-f600.png","2026-05-13T14:10:25.488549+00:00",{"id":72,"slug":73,"title":74,"cover_image":75,"image_url":75,"created_at":76,"category":27},"0af0a4b2-b0a1-4a52-8fe9-1328bde87c8e","why-ethereum-treasury-buying-is-a-bad-bet-zh","為什麼 Ethereum Treasury Buying 正在變成一筆差勁的長…","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1778386236909-ytls.png","2026-05-10T04:10:21.784208+00:00",{"id":78,"slug":79,"title":80,"cover_image":81,"image_url":81,"created_at":82,"category":27},"ab3ef302-99ee-40b3-b2d0-4b67a9049ec4","yakovenko-warns-ai-could-crack-pqc-wallets-zh","Yakovenko 警告：AI 可能破解 PQC 錢包","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1778170266863-wnnh.png","2026-05-07T16:10:41.097774+00:00",[84,89,94,99,100,105,110,115,120,125],{"id":85,"slug":86,"title":87,"created_at":88},"e1b4b518-f86b-410c-8c82-8cfb787ff2ef","moonpay-open-wallet-standard-ai-payments-zh","MoonPay 推 OWS，瞄準 AI 付款","2026-03-28T03:08:33.379969+00:00",{"id":90,"slug":91,"title":92,"created_at":93},"e72bae29-ddbd-437b-aaa4-cd662605394b","next-gen-crypto-simulators-ai-web3-training-zh","新一代加密模擬器更聰明了","2026-04-01T09:36:33.917023+00:00",{"id":95,"slug":96,"title":97,"created_at":98},"b8e39b58-6b9d-4714-92d3-26df18a3e0f4","rtk-cuts-claude-code-token-spend-zh","RTK 讓 Claude Code 少燒 Token","2026-04-01T10:24:29.259497+00:00",{"id":4,"slug":26,"title":5,"created_at":22},{"id":101,"slug":102,"title":103,"created_at":104},"00668dea-9f0e-4019-b861-03817d5a8877","how-web3-marketing-changed-in-2026-zh","2026 Web3 行銷怎麼變了","2026-04-02T01:36:34.973322+00:00",{"id":106,"slug":107,"title":108,"created_at":109},"e7992274-42ee-40bc-bb05-97250098c56c","ai-agentic-defi-web3-grants-march-2026-zh","AI、Agentic DeFi 與 Web3 補助案","2026-04-02T05:51:36.857954+00:00",{"id":111,"slug":112,"title":113,"created_at":114},"5cef810b-af3d-467a-8b41-627769eca895","why-crypto-is-fixated-on-ai-agents-zh","為何加密圈盯上 AI Agent","2026-04-02T05:54:28.919864+00:00",{"id":116,"slug":117,"title":118,"created_at":119},"d30e6203-d522-41a1-b529-fcf4499cd985","web3-explained-what-it-is-why-it-matters-zh","Web3 是什麼，為何重要","2026-04-02T06:15:32.580114+00:00",{"id":121,"slug":122,"title":123,"created_at":124},"f29e65ae-64df-463b-ba22-afd9dcbd0f8f","trust-wallet-agent-kit-ai-trade-25-chains-zh","Trust Wallet 讓 AI 幫你交易","2026-04-02T06:27:33.183404+00:00",{"id":126,"slug":127,"title":128,"created_at":129},"91022b4c-b53e-4c18-abfe-914a8eca6e28","blockchain-in-ai-real-use-cases-zh","區塊鏈加 AI，真實落地在哪裡","2026-04-02T06:30:44.026286+00:00"]