Mấy hôm nay fb ở Việt nam đang rộ lên trò chơi DeepFake - một trò tưởng chừng như vui thôi, không quạu nhưng thực chất có thể làm đảo lộn thế giới.
Đây là một công nghệ trí tuệ nhân tạo A.I, trên nguyên lý học sâu (Deep Learning) nhưng tập trung vào lĩnh vực nhái (Fake) con người.
Nhái giọng thì dễ và có lâu rồi, nhái ảnh thì gần đây có nhiều bước tiến, nhưng nay nhái chuyển động - điệu cười, bước đi, cái đánh hông, và cả cái đá lông nheo chết người nữa - thì thật hết chịu nổi.
Rất nhiều chuyện không thể ko quạu đã xảy ra: như là clip Obama chửi mắng Trump trước kỳ bầu cử, hay Trump nhận nợ với Putin về nợ quốc gia..., và nhiều nhất là các ngôi sao nổi tiếng bị làm giả trong các phim cấp 3, hoặc sếp nhớn bị giả giọng và hình ảnh trên các nền tảng Video streaming để họp và chỉ đạo online ra lệnh ...chuyển tiền!
Nguyên lý của máy học, là chúng ta càng tạo ra, càng khen ngợi / chửi bới (có phản ứng như là like/dislike, mở xem, bình luận...) thì fake sẽ càng deep.
Và rất có thể một ngày mai thức dậy, bạn thấy một clip của chính mình đang làm một việc bẩn thỉu mà bạn ko thể thanh minh hoặc được vạ thì má đã sưng.
Vậy nên vui hay ko nên, tuỳ bạn!
#TeamSharkHung
同時也有2部Youtube影片,追蹤數超過75萬的網紅志祺七七 X 圖文不符,也在其Youtube影片中提到,✔︎ 成為七七會員(幫助我們繼續日更,並享有會員專屬福利):http://bit.ly/shasha77_member ✔︎ 體驗志祺七七文章版:https://blog.simpleinfo.cc/shasha77 ✔︎ 購買黃臭泥周邊商品: https://reurl.cc/Ezkbma 💛 ✔...
「deepfake online」的推薦目錄:
- 關於deepfake online 在 Facebook 的最讚貼文
- 關於deepfake online 在 Lee Hsien Loong Facebook 的最佳貼文
- 關於deepfake online 在 AppWorks Facebook 的最佳貼文
- 關於deepfake online 在 志祺七七 X 圖文不符 Youtube 的精選貼文
- 關於deepfake online 在 暗網仔 2.0 Youtube 的精選貼文
- 關於deepfake online 在 Here's How Easy It Is To Create Deepfake Videos Online 的評價
- 關於deepfake online 在 The future of online trust (and why Deepfake is advancing it) 的評價
- 關於deepfake online 在 deepfake maker online free的推薦與評價,YOUTUBE和網紅 ... 的評價
deepfake online 在 Lee Hsien Loong Facebook 的最佳貼文
Technology cuts both ways. While it has undoubtedly made our lives easier, it has also enabled new and insidious types of criminal activity – cyber crime. As Artificial Intelligence becomes more sophisticated, it is becoming more difficult to believe what you see or hear. First, there were fake ads using my photo. Now, criminals are faking voices to trick their victims.
In the first so-called ‘AI heist’, criminals used voice-imitating AI to mimic a CEO’s voice, to trick his MD into transferring several hundred thousand dollars of company funds to them.
Such AI programs can be easily trained to mimic voices and speech patterns from public sources like interviews and YouTube videos. Some programs are available for free, so anyone can download and use deepfake software to recreate anyone's voice.
We have to adapt quickly. Tech companies are developing ways to detect deepfake audio and video. Meanwhile, we should be discerning about what we come across online, whether or not it seems legitimate. Always be careful. If in doubt, check first before you proceed. You can call the National Crime Prevention Council's anti-scam helpline (1800-722-6688) for advice, or visit the scam alert website www.scamalert.sg. – LHL
deepfake online 在 AppWorks Facebook 的最佳貼文
【Where's the Lie?】
Deepfake (a portmanteau of "deep learning" and "fake") is a technique for human image synthesis based on artificial intelligence. It is used to combine and superimpose existing images and videos onto source images or videos using a machine learning technique.
Deepfake technology is going to get better and better. Recently, there has been a rise of deepfakes on social media, for example the one of House Speaker Nancy Pelosi in which it appears she is slurring her speech. Facebook didn't take down the video, which prompted the creation of a deepfake of Facebook founder Mark Zuckerberg saying how he wanted to take over the world, which was uploaded on Instagram.
People and companies have started to worry about the spread of deepfakes, and fake videos and pictures, and with good reason. One such company is Adobe, which recently shared new research in collaboration with UC Berkeley scientists that uses machine learning to automatically detect when images of faces have been manipulated.
While it's still in the research stage, it's promising that there's technology out there to combat manipulative deepfakes online. However, these kinds of tools can only go so far to stop harmful content and media. Knowing something is fake is only half the battle, as we've seen with fake news spreading. Let's hope that as we see a rise in deepfakes, we see other technologies and startups that can counter the rise in manipulation.
Want to be more involved in an AI community? Apply now to be a part of AppWorks Accelerator >> http://bit.ly/2IQX45l
- Natalie Feng Lin, Analyst.
deepfake online 在 志祺七七 X 圖文不符 Youtube 的精選貼文
✔︎ 成為七七會員(幫助我們繼續日更,並享有會員專屬福利):http://bit.ly/shasha77_member
✔︎ 體驗志祺七七文章版:https://blog.simpleinfo.cc/shasha77
✔︎ 購買黃臭泥周邊商品: https://reurl.cc/Ezkbma 💛
✔︎ 訂閱志祺七七頻道: http://bit.ly/shasha77_subscribe
✔︎ 追蹤志祺IG :https://www.instagram.com/shasha77.daily
✔︎ 來看志祺七七粉專 :http://bit.ly/shasha77_fb
✔︎ 如果不便加入會員,也可從這裡贊助我們:https://bit.ly/support-shasha77
(請記得在贊助頁面留下您的email,以便我們寄送發票。若遇到金流問題,麻煩請聯繫:service@simpleinfo.cc)
#韓國 #N號房 #您已登入N號房
各節重點:
00:00 前導
01:25 「會員福利介紹」廣告段落
02:35 事情的開端
03:24 意外發現N號房
04:37 N號房不是大家都進得去?
05:26 N號房裡到底有什麼?
06:37 蒐證讓她們遇上前所未有的恐懼?
07:51 報導N號房,反被加害者肉搜?
08:57 我們的觀點
10:26 提問
10:42 結尾
【 製作團隊 】
|企劃:KT
|腳本:KT、羊羊
|編輯:土龍
|剪輯後製:絲繡
|剪輯助理:歆雅/珊珊
|演出:志祺
——
【 本集參考資料 】
→《您已登入N號房》,追蹤團火花著,時報出版
→楊虔豪專文導讀/從金智恩到N號房:南韓 MeToo與「不完美受害者」的勇氣|轉角國際:https://bit.ly/3slzdkA
→葉靜倫/一起讓星火燎原──讀《您已登入N號房》:https://bit.ly/3g73L4M
→獨家專訪》韓國N號房事件吹哨者「追蹤團火花」 女大生臥底性犯罪地獄實錄「好像,只有我們可以救她。」|今周刊:https://bit.ly/3AM16oU
→李慈媛/每100位男性就有1個會員?南韓「N號房事件」的虛與實|鳴人堂:https://bit.ly/3AMBZ5k
【 延伸閱讀 】
→南韓「N號房事件」殘酷一年後?屠殺靈魂的現場直播間|轉角國際:https://bit.ly/3k3UhYT
→有錢就能散布別人裸照,誰讓加害者有恃無恐?|報導者:https://bit.ly/3k5gR3B
→私密照外流之痛:誰讓我淪為「國家認證猥褻物」|公視新聞實驗室:https://bit.ly/3CXq46s
→臉被偷走之後:無法可管的數位性暴力?台灣Deepfake 事件獨家調查|鏡週刊:https://bit.ly/3AFF7zY
→「影片裡是你嗎?」私密影像被公開在網路上,該怎麼辦?:https://bit.ly/3g7Dlj2
→人人都可能成為A片主角?|志祺七七:https://bit.ly/3k7rL98
→一篇報導讓Pornhub壓力山大?|志祺七七:https://bit.ly/3g6beRh
→GMA News Online, The Children Are All Grown Up Now:https://bit.ly/3yU3klH
→為什麼,我們會指責受害者,而非加害者呢?|志祺七七:https://bit.ly/3yUMmDH
→復仇式色情你我都有份?|志祺七七:https://bit.ly/37QhesY
\每週7天,每天7點,每次7分鐘,和我們一起了解更多有趣的生活議題吧!/
🥁七七仔們如果想寄東西關懷七七團隊與志祺,傳送門如下:
106台北市大安區羅斯福路二段111號8樓
🟢如有引用本頻道影片與相關品牌識別素材,請遵循此規範:http://bit.ly/shasha77_authorization
🟡如有業務需求,請洽:hi77@simpleinfo.cc
🔴如果影片內容有誤,歡迎來信勘誤:hey77@simpleinfo.cc
deepfake online 在 暗網仔 2.0 Youtube 的精選貼文
現在有一個聖誕Promo: 只要按Description ‘Deepwebkid’ 的link NordVPN申請有兩折優惠! 入會3年更送多你四個月的plan和一個價值$195美元的密碼儲存機.
請使用以下連結: https://nordvpn.com/deepwebkid
優惠碼: deepwebkid
會員:
https://www.youtube.com/channel/UC8vabPSRIBpwSJEMAPCnzVQ/join
Jeff電單車VLOG: https://www.youtube.com/c/otobaiheathk?fbclid=IwAR0nXGa7C9MjRhJx2TskwgTjQjrQlYh5GY_khrSC0ilMvwU5WPfIunsVImI
Instagram: https://www.instagram.com/dw_kid12/
Facebook: https://www.facebook.com/deepwebkid/?modal=admin_todo_tourSubscribe:
https://www.youtube.com/channel/UC8vabPSRIBpwSJEMAPCnzVQ?sub_confirmation=1
Deepfake換臉程式的黑暗秘密: 人類未來的危機
大家有沒有覺得這段Obama的演講有種很奇怪的感覺啊? 其實該片段是一年前導演Jordan Peele用一個名Deepfake的恐怖AI科技Key了美國總統的樣子上去的. Deepfake是能把任何人的樣子放在任何身體上. 用這種科技出ham片!楊幂中國deepfake事件! 甚至未來政治危機! 之後一一會講.
在危機處處的網絡世界我是用NordVPN保護自己的. 如果你們不想黑客入侵,個人資料被盜用NordVPN一向也是我上暗網的保護罩. 上去食人族暗網’wun食’? NordVPN! 到北韓網站看. NordVPN! Nordvpn無論Android iphone也可以用, 在機場, coffee shop, 甚至到中國大陸也行, 讓我本身住加拿大這個事實不容易被發現.
現在有一個聖誕Promo: 只要gum Description ‘Deepwebkid’ 的link NordVPN申請有兩折優惠! 入會3年更送多你四個月的plan和一個價值$195美元的密碼儲存機.
This is coming from my heart. Especially those with internet businesses like myself, this service not only secured our online identity but also our livelihood and safety. And this plan works in 6 simaltanoeus locations, meaning you and your family can all be protected from the dangers of the deep web.
Deepfake源自於一個名Interval Research corporation的網路科技研究中心. 1997年研究出Video rewrite program, 初步能夠將影片?面口型的畫面配合上後期加上的對白. 在Hollywood電影後製變成很重要的科技. 但要到20年之後Deepfake才得到主流關注. 2017年我們看到的 ‘Synthesizing Obama’ 網上流傳. 2017年尾這種科技才由授權人士掉到一般人的手中. 例如這個名 ‘fakeapp’ 的應用程式. 結果問題就開始了.
今天就講一下deepfake危機會出現的三個地方.
色情
2017年尾名 ‘deepfakes’ 的一位reddit討論區用戶上載了第一個自製Deepfake. 之後luk jook有各個女明星的deepfake出現. 女星的樣子放在A片女星的身體上. 高fung時一天數千條這些影片放上討論區.
但如果你只是普通的女生, 一醒來發現無數張有你樣子的色情照片和影片, 你又如何接受呢?
法律學生Noelle martins 17歲的時候發現網上有大量photoshop她的樣子的色情照片. 之後的5年, Noelle與各大的色情網站進行一個gim gooy的戰爭. 除即發現她身邊的朋友甚至自己的妹妹也成為這些Deepfake網站的受害者.
這個熱潮導致兩極分化. 道德上該事件一定是重大問題. 但另一邊學者認為色情行業一向都是推動新科枝的重要元素.
這其實sip及西方社會一個敏感的A片類型: 復仇色情片. 簡單講, 復仇色情片是一些ja男與女友分手後把她的裸照或親密片放給大家看的色情片內容.
Deepfake卻令到沒有拍過任何東西的女生也可能成為 ‘復仇’ 目標. Noelle martins就是一個越反抗越成為目標的例子.
楊冪
國內好多科技上的潮流也是來自外國. Deepfake也不列外. 這張照片曾在國內引發很大的關注. 大家覺得有問題嗎?
這張内地女演員楊冪的造型照來自於TVB 1994年射雕英雄傳 ‘黃蓉’ 的角色. 但1994年的楊冪只有8歲, 其實這是一個deepfake, 原本的造型是朱茵. 但真實程度讓影片出現在 ‘B站’ Bilibili後, 這種 ‘合成技術’ 在國內全面禁止.
Political event
外國明星樣子被deepfake惡搞卻常常發生. 演員Nicolas cage的樣子出現最多. Wonder Woman Gal Gadot的事件最為人熟悉.但看看這個片段(show hilary Clinton Saturday night live deepfake)
而這是美國總統Donald Trump. 被一名演員在模仿.
這科技成為政治工具一向是公眾最大yuw leuy. Shane Dawson的 ‘conspiracy theories with Shane Dawson’ 對deepfake也將這當作最大恐懼.
-2020 election
但將近2020年美國election, 一些不懂政治的人好可能會因爲看到領導者假的片段導致大gook有所改變. 正式加強一場名 ‘Cyberwar’ 的東西. Hacking, computer virus有危險性之外, 用media一些假的慨念到大眾的思想?面. 又名 ‘likewar’
...即是騙 ‘like’ 吧!
What’s being done to stop it
兩個月前Google發送了幾千條Deepfake影片給researchers. 幫助建立AI能查出那一些影片有deepfake成份. 美國有兩個州份Virginia和Texas已經法例禁止deepfake成人電影的發行. 同時雖然Deepfake本身在整個美國不是犯法, 用一條假的影片有意做詐騙行為是犯法的.
同月facebook創辦人mark zuckerberg國會大樓?面答應會嚴格監督Facebook feed 上面的新聞.
deepfake online 在 The future of online trust (and why Deepfake is advancing it) 的美食出口停車場
The danger of untrustworthy AI often culminates with Deepfake, perceived as unprecedented threat for democracies and online trust, through its potential to ... ... <看更多>
deepfake online 在 Here's How Easy It Is To Create Deepfake Videos Online 的美食出口停車場
You should not believe everything you see, because advancements in software have made it easy to create deepfake videos. ... <看更多>