Kualitas bukti 4.5/5
Skor tinjauan delapan dimensi terhadap rubrik kualitas . Setiap dimensi dinilai 1–5.
- D1 Dasar sumber
- 3/5
- D2 Otoritas sumber
- 5/5
- D3 Aritmetika
- 5/5
- D4 Ketidakpastian
- 5/5
- D5 Cakupan
- 4/5
- D6 Prosa
- 5/5
- D7 Kejujuran persepsi
- 4/5
- D8 Kelengkapan peringatan
- 5/5
Laporan Kejahatan Internet FBI 2025 menandai pertama kalinya badan tersebut memisahkan penipuan berbasis AI sebagai kategori tersendiri, mencatat 22.364 pengaduan dan hampir $893 juta kerugian yang dilaporkan. Kloning suara, video deepfake, dan dokumen yang dihasilkan AI muncul dalam penipuan investasi, peniruan keadaan darurat keluarga, penipuan asmara, dan — dalam pola yang lebih baru — penipuan wawancara kerja di mana kandidat atau pemberi kerja menggunakan pertukaran wajah secara real-time. Orang dewasa yang lebih tua menanggung bagian yang tidak proporsional: $352 juta dari total $893 juta. Ini hanya angka yang dilaporkan; FBI mengakui bahwa pengaduan IC3 hanya menangkap sebagian kecil dari viktimisasi aktual, dan perkiraan independen mengenai pelaporan kurang berkisar dari 5x hingga 10x.
Lintasan pertumbuhan adalah yang membedakan penipuan deepfake dari kategori kejahatan siber yang sudah mapan. Sumsub, platform verifikasi identitas yang memproses jutaan pemeriksaan secara global, melaporkan bahwa deepfake naik dari 7% dari semua upaya penipuan yang terdeteksi pada 2024 menjadi 11% pada 2026 — hampir dua kali lipat dalam dua tahun. Teknologi yang mendasarinya berada pada kurva biaya-kinerja yang curam: kloning suara yang membutuhkan beberapa menit sampel audio pada 2023 kini bekerja andal dengan klip tiga detik, dan alat pertukaran wajah real-time berjalan pada perangkat keras konsumen. Hambatan masuk bagi calon penipu turun lebih cepat daripada kemampuan alat pertahanan untuk beradaptasi, itulah sebabnya tahun pertama pelacakan FBI sudah menghasilkan angka kerugian sembilan digit.
Tidak ada probabilitas seumur hidup yang dapat diandalkan yang ditawarkan di sini karena ancaman terlalu baru dan bergerak terlalu cepat untuk tingkat dasar yang stabil. 22.364 pengaduan pada 2025 mewakili kategori yang secara harfiah tidak ada dalam pelaporan IC3 tahun sebelumnya. Mengekstrapolasi angka seumur hidup per orang dari satu tahun data parsial, selama periode pertumbuhan kapabilitas eksponensial, akan menghasilkan angka yang sekaligus tepat dan salah. Yang didukung data adalah klaim arah: penipuan deepfake sudah merupakan risiko keuangan material bagi individu biasa, bukan hanya korporasi atau tokoh publik, dan persepsi publik belum menyusul. Entri ini ditandai sebagai diremehkan sesuai dengan itu.
Fakta terkait
Tidak ada tingkat per orang yang andal untuk penipuan identitas deepfake. Sinyalnya kuat dan terus tumbuh, tetapi penyebutnya tidak jelas dan pelacakan baru dimulai belakangan ini, sehingga setiap probabilitas tunggal akan terlalu dini. Ancamannya nyata; angkanya belum dapat diukur.
Buku besar klaim
Setiap angka di bawah ini adalah apa yang dilaporkan masing-masing sumber, dengan kutipan kata demi kata yang kami andalkan dan bagaimana kami sampai pada angka kami. Klik tautan mana saja untuk memverifikasi langsung.
4/4 sumber diverifikasi secara independen kata demi kata terhadap sumber terkutip
-
[1] Federal Bureau of Investigation — Cryptocurrency and AI Scams Bilk Americans of Billions Terverifikasi
Cryptocurrency and AI Scams Bilk Americans of Billions- Statistik
22,364 AI-related complaints in 2025 with losses of nearly $893 million; voice cloning and deepfake videos used in investment schemes and family-emergency impersonations- Kutipan
“"For the first time in its nearly 25-year history, the IC3 report features a section on artificial intelligence, which accounts for 22,364 complaints, costing Americans nearly $893 million." ”
- Data sumber dari
- 2026-04-23
- Diakses
- 2026-04-24 · salinan arsip
- Verifikasi
- Kutipan diambil ulang secara independen dan dikonfirmasi kata demi kata terhadap sumber terkutip selama audit pendasaran kami.
- Perhitungan
- The 2025 IC3 Annual Report is the first to break out AI-enabled fraud as a distinct category. The $893 million figure covers complaints where AI tools — including voice cloning, deepfake video, and AI-generated documents — were identified as enabling the fraud. Older adults accounted for $352 million of those losses. The FBI notes that actual losses are likely far higher because IC3 captures only a fraction of incidents (the FTC estimates IC3 complaint rates at roughly 10-15% of actual victimisation). Even with a conservative 5-10x underreporting multiplier, the annualised figure would be $4.5-9 billion — but the per-person probability remains impossible to pin down because (a) the denominator is unclear (US adults? global internet users?), (b) the technology is improving faster than annual reporting can track, and (c) 2025 is the first year with any dedicated tracking. This is why the entry uses no_reliable_estimate.
-
[2] Sumsub — Fraud Trends 2026: AI Scams, Deepfakes, and Emerging Threats Terverifikasi
Fraud Trends 2026: AI Scams, Deepfakes, and Emerging Threats- Statistik
Deepfakes accounted for 7% of all fraud attempts globally in 2024, rising to 11% in 2026; a 4x increase in deepfake detections from 2023 to 2024- Kutipan
“"In 2026, AI scams are everywhere, with deepfakes now accounting for 11% of global fraudulent activity." ”
- Data sumber dari
- 2026-01-15
- Diakses
- 2026-04-24 · salinan arsip
- Verifikasi
- Kutipan diambil ulang secara independen dan dikonfirmasi kata demi kata terhadap sumber terkutip selama audit pendasaran kami.
- Perhitungan
- Sumsub is an identity verification platform that processes millions of verification checks globally and publishes an annual fraud report based on its operational data. The 7% → 11% trajectory from 2024 to 2026 is directionally consistent with the FBI's first-time inclusion of an AI section in its 2025 report. The 4x year-over-year increase in deepfake detections (2023 to 2024) reflects both improved detection tooling and genuine growth in deepfake usage. The Sumsub data covers identity verification contexts (onboarding, KYC) rather than all fraud, so it likely understates the share of deepfakes in social-engineering scams (romance fraud, family-emergency calls) where no verification check is attempted. The rapid growth rate is precisely what makes a stable lifetime probability estimate unreliable — any number computed from 2025 data would be stale before publication.
- Independensi
- Sumsub's fraud detection data is independently collected from its own verification platform, distinct from FBI IC3 complaint-based reporting. The two datasets measure different things (attempted identity fraud at verification vs reported financial losses) and corroborate each other directionally.
-
[3] Deloitte Center for Financial Services — Generative AI is expected to magnify the risk of deepfakes and other fraud in banking Terverifikasi
Generative AI is expected to magnify the risk of deepfakes and other fraud in bankingSee all 2 Likelier entries citing this source →
- Statistik
Deloitte's 2024 projection: gen-AI-enabled fraud losses in the US could reach $40 billion by 2027, up from $12.3 billion in 2023 — a 32% compound annual growth rate- Kutipan
“"gen AI could enable fraud losses to reach US$40 billion in the United States by 2027, from US$12.3 billion in 2023, a compound annual growth rate of 32%." ”
- Data sumber dari
- 2024-05-29
- Diakses
- 2026-06-14 · salinan arsip
- Verifikasi
- Kutipan diambil ulang secara independen dan dikonfirmasi kata demi kata terhadap sumber terkutip selama audit pendasaran kami.
- Perhitungan
- Deloitte's Center for Financial Services built this projection by assigning generative-risk scores to 26 IC3 complaint categories and modelling forward growth; the conservative variant lands nearer $22 billion. It is a forward-looking industry projection, not an observed base rate, and it does not yield a per-person lifetime probability — it forecasts an aggregate US dollar loss. It is included here to bound the trajectory and corroborate the FBI's first-year nine-figure AI-fraud loss: an independent analyst forecasts roughly 3.25x growth ($12.3B to $40B) over four years, consistent with the steep curve that makes any stable per-person estimate premature. This reinforces, rather than resolves, the no_reliable_estimate flag.
-
[4] SecureWorld — FBI: AI-Enabled Fraud Topped $893M in 2025 — Real Toll Likely Far Higher Terverifikasi
FBI: AI-Enabled Fraud Topped $893M in 2025 — Real Toll Likely Far Higher- Statistik
FBI acknowledges actual AI fraud losses likely far exceed $893M due to chronic underreporting; voice deepfakes used in job interview fraud cost victims $13 million in 2025- Kutipan
“"The FBI documented widespread use of voice spoofing and video deepfakes during online job interviews in 2025, with victims reporting losses of approximately $13 million." ”
- Data sumber dari
- 2026-04-24
- Diakses
- 2026-04-24 · salinan arsip
- Verifikasi
- Kutipan diambil ulang secara independen dan dikonfirmasi kata demi kata terhadap sumber terkutip selama audit pendasaran kami.
- Perhitungan
- SecureWorld's analysis of the 2025 IC3 report highlights the underreporting problem that prevents reliable probability estimation. The $13 million job-interview-deepfake figure is a narrow subcategory illustrating how deepfake fraud extends beyond the stereotypical wire-transfer scam into employment, romance, and investment contexts. The article notes that the FBI itself considers reported figures a significant undercount. This corroborates the decision to use no_reliable_estimate: the signal is strong (deepfake fraud is large and growing), but the noise in the denominator (who is exposed? how often?) makes a per-person probability unreliable.