{
  "id": 122168,
  "title": "Why don't we make Deepfake model as private?",
  "url": "/competitions/deepfake-detection-challenge/discussion/122168",
  "author_name": "",
  "post_date": "2019-12-18T07:43:53.678977600Z",
  "votes": null,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Acknowledgment\n- This discussion is not related to the DFDC nor Deepfake model.\n- This paragraph would be like a devil's advocate. you can say anything about this. </p>\n\n<p>I attended the seminar about Deepfake on ICCV 2019 and I was wondering that Deepfake (and other derivative models such as Face2Face and FaceSwap)  model should be secretly published to researchers and should not be published to the public.\nEven though the right to know is important but this model has a high potential for adverse effects on society. \nWe know that the Deepfake model is started by an unknown user by malicious purpose. So we suppose that even though we build the best detection model of Deepfake, still there is some possibility about the adversarial model to fool the detection model will come out. \nSorry for poor writing but I really want to know the opinions about\n 1. Make it private or limit the access about Deepfake paper (or derivative of Deepfake)\n 2. Make it private or limit the access about Detection of Deepfake (or derivative of Deepfake)</p>",
  "messages": [
    {
      "id": "697630",
      "postDate": "12/18/2019 07:43:53",
      "content": "<p>Acknowledgment\n- This discussion is not related to the DFDC nor Deepfake model.\n- This paragraph would be like a devil's advocate. you can say anything about this. </p>\n\n<p>I attended the seminar about Deepfake on ICCV 2019 and I was wondering that Deepfake (and other derivative models such as Face2Face and FaceSwap)  model should be secretly published to researchers and should not be published to the public.\nEven though the right to know is important but this model has a high potential for adverse effects on society. \nWe know that the Deepfake model is started by an unknown user by malicious purpose. So we suppose that even though we build the best detection model of Deepfake, still there is some possibility about the adversarial model to fool the detection model will come out. \nSorry for poor writing but I really want to know the opinions about\n 1. Make it private or limit the access about Deepfake paper (or derivative of Deepfake)\n 2. Make it private or limit the access about Detection of Deepfake (or derivative of Deepfake)</p>",
      "rawMarkdown": "Acknowledgment\n- This discussion is not related to the DFDC nor Deepfake model.\n- This paragraph would be like a devil's advocate. you can say anything about this. \n\nI attended the seminar about Deepfake on ICCV 2019 and I was wondering that Deepfake (and other derivative models such as Face2Face and FaceSwap)  model should be secretly published to researchers and should not be published to the public.\nEven though the right to know is important but this model has a high potential for adverse effects on society. \nWe know that the Deepfake model is started by an unknown user by malicious purpose. So we suppose that even though we build the best detection model of Deepfake, still there is some possibility about the adversarial model to fool the detection model will come out. \nSorry for poor writing but I really want to know the opinions about\n 1. Make it private or limit the access about Deepfake paper (or derivative of Deepfake)\n 2. Make it private or limit the access about Detection of Deepfake (or derivative of Deepfake)",
      "votes": null
    },
    {
      "id": "699533",
      "postDate": "12/20/2019 14:50:27",
      "content": "<p>Open-sourse makes technology improving faster.</p>",
      "rawMarkdown": "Open-sourse makes technology improving faster.",
      "votes": null
    },
    {
      "id": "949348",
      "postDate": "07/28/2020 15:20:52",
      "content": "<p>There is no way to stop technology evolving, you can make models private, there will be others coming out... That is like saying \"stop making Artificial Intelligence smart, they will take over the world\". The tech is here, all we can do is detect the fakes but we have to move forward, not backward.</p>",
      "rawMarkdown": "There is no way to stop technology evolving, you can make models private, there will be others coming out... That is like saying \"stop making Artificial Intelligence smart, they will take over the world\". The tech is here, all we can do is detect the fakes but we have to move forward, not backward.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 699533,
      "author_name": "diegojohnson",
      "author_url": "",
      "post_date": "12/20/2019 14:50:27",
      "content": "<p>Open-sourse makes technology improving faster.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 949348,
      "author_name": "alincijov",
      "author_url": "",
      "post_date": "07/28/2020 15:20:52",
      "content": "<p>There is no way to stop technology evolving, you can make models private, there will be others coming out... That is like saying \"stop making Artificial Intelligence smart, they will take over the world\". The tech is here, all we can do is detect the fakes but we have to move forward, not backward.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "697630": "Acknowledgment\n- This discussion is not related to the DFDC nor Deepfake model.\n- This paragraph would be like a devil's advocate. you can say anything about this. \n\nI attended the seminar about Deepfake on ICCV 2019 and I was wondering that Deepfake (and other derivative models such as Face2Face and FaceSwap)  model should be secretly published to researchers and should not be published to the public.\nEven though the right to know is important but this model has a high potential for adverse effects on society. \nWe know that the Deepfake model is started by an unknown user by malicious purpose. So we suppose that even though we build the best detection model of Deepfake, still there is some possibility about the adversarial model to fool the detection model will come out. \nSorry for poor writing but I really want to know the opinions about\n 1. Make it private or limit the access about Deepfake paper (or derivative of Deepfake)\n 2. Make it private or limit the access about Detection of Deepfake (or derivative of Deepfake)",
    "699533": "Open-sourse makes technology improving faster.",
    "949348": "There is no way to stop technology evolving, you can make models private, there will be others coming out... That is like saying \"stop making Artificial Intelligence smart, they will take over the world\". The tech is here, all we can do is detect the fakes but we have to move forward, not backward."
  },
  "source": "meta"
}