AI 模型识别伪造网址，准确率最高达 99%

September 16, 2026 AI model spots spoofed website addresses with up to 99% accuracy by David Bradley, Inderscience edited by Lisa Lock, reviewed by Andrew Zinin Lisa Lock Scientific Editor Meet our editorial team Behind our editorial process Andrew Zinin Chief Editor Meet our editorial team Behind our editorial process Editors' notes This article has been reviewed according to Science X's editorial process and policies. Editors have highlighted the following attributes while ensuring the content's credibility: fact-checked peer-reviewed publication trusted source proofread The GIST Add as preferred source Credit: Pixabay/CC0 Public Domain A machine-learning system can identify fake, or spoofed, website addresses, according to research published in the International Journal of Electronic Security and Digital Forensics. The work offers a new tool to protect users against phishing attacks, in which attackers direct someone to a deceptive website to steal information such as bank logins or personal and private data. The team combined two deep-learning techniques: a convolutional neural network (CNN), which can identify patterns in data, and a long short-term memory (LSTM) network, which can retain information about sequences in that data. This hybrid CNN-LSTM model achieved accuracy rates of almost 99% on the UCL standard test dataset and almost 97% on the PhishTank dataset. The findings highlight the potential for automated AI systems to protect people using online commerce, government services and other activities. Phishing attacks remain particularly difficult to address as a cybercrime problem because they often exploit human behavior and weaknesses rather than a technical loophole. If a message is sufficiently convincing and a website realistic, vulnerable or even distracted users might disclose sensitive information without realizing they have been duped. A system that accurately flags spoofed addresses before that happens would be invaluable. For the small percentage of addresses it falsely flags, the inconvenience would be negligible compared with what is not lost by blocking spoofed sites. Publication details B.V. Santhosh Krishna et al, A hybrid deep learning method for URL spoofing in websites, International Journal of Electronic Security and Digital Forensics (2026). DOI: 10.1504/ijesdf.2026.156138 Journal information: International Journal of Electronic Security and Digital Forensics Key concepts Cybersecurity frameworksTrustworthy machine learning Provided by Inderscience Who's behind this story? Lisa Lock BA art history, MA material culture. Former museum editor, paramedic, and transplant coordinator. Editing for Science X since 2021. Full profile → Andrew Zinin Master's in physics with research experience. Long-time science news enthusiast. Plays key role in Science X's editorial success. Full profile → Citation: AI model spots spoofed website addresses with up to 99% accuracy (2026, September 16) retrieved 16 September 2026 from https://techxplore.com/news/2026-09-ai-spoofed-website-accuracy.html This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no part may be reproduced without the written permission. The content is provided for information purposes only. Smarter machine-learning models can improve phishing website detection Feb 10, 2026 PhishLumos maps phishing infrastructure and finds 190,000 URLs in six months Jun 13, 2026 Research offers novel approach for combating phishing attacks Sep 12, 2023 A new AI-based tool to detect DDoS attacks Feb 28, 2023 A model that can help inexperienced users identify phishing emails Apr 19, 2022 15-minute training helps people spot AI-generated faces Sep 3, 2026 Why colder temps can be better for aircraft deicing systems 42 minutes ago Tiny polymer devices mimic neuron firing, pointing toward energy-efficient edge computing 1 hour ago 'We're losing control': AI pioneer urges nuclear-style safeguards for autonomous agents 1 hour ago Steelmaking waste could store solar energy, with electric arc furnace slag leading early tests 1 hour ago Manuscripts-turned AI agents can now 'talk' to each other and make new discoveries 1 hour ago It bends, but does not break: A new sustainable treatment for shaping wood 1 hour ago