Interview with Sam Gregory of WITNESS about synthetic media and disinformation

  • home
  • /
  • Blog Posts
  • News
  • /
  • Interview with Sam Gregory of WITNESS about synthetic media and disinformation
By on January 19th, 2021 in Blog Posts, News

Sam Gregory is Program Director of WITNESS, an organisation that works with people who use video to document human rights issues. WITNESS focuses on how people create trustworthy information that can expose abuses and address injustices. How is that connected to deepfakes?

Deutsche Welle’s Julia Bayer conducted an interview with the Program Director of witness.org, Sam Gregory. The interview was originally published on the DIGGER project website and conducted in the context of this Google DNI-funded project in which DW and ATC also participate (together with Fraunhofer IDMT). It also includes video sequences of the interview featuring what Sam had to say on the matter.

Here, we re-publish the (edited and condensed) interview with kind permission of the DIGGER project and the individuals involved. In relation to the original on the DIGGER project website, some minor edits were made by Jochen Spangenberg for publication on the WeVerify website.

Topical focus of what follows: the development and challenges of new ways to create mis- and disinformation, specifically those making use of artificial intelligence. This includes discussing the impact of shallow- and deepfakes, and what the essential questions are with the development of tools for detection of such synthetic media.

About WITNESS: WITNESS is an organisation that works with people who use video to document human rights violations and related matters. The NGO furthermore focuses on how people can create trustworthy information that can expose abuses and address injustices.

*****

JULIA BAYER: Sam, what is your definition of a deepfake?

SAM GREGORY: I use a broad definition of a deepfake. I use the phrase synthetic media to describe the whole range of ways in which you can manipulate audio or video with artificial intelligence.

We look at threats, and in our search for solutions we look at how you can change audio, how you can change faces and how you can change scenes by for example removing objects or adding objects more seamlessly.

What is the difference to shallowfakes?

We use the phrase shallowfake in contrast to deepfake to describe what we have seen for the past decade at scale, which is people primarily miscontextualizing videos like claiming a video is from one place when it is actually from another place. Or claiming it is from one date when it is actually from another date. Also, when people do deceptive edits of videos or do things you can do in a standard editing process, like slowing down a video, we call it a shallowfake.

The impact can be exactly the same, but I think it’s helpful to understand that deepfakes can create these incredibly realistic versions of things that you haven’t been able to do with shallowfakes. For example, the ability to make someone look like they’re saying something or to make someone’s face appear to do or say something that they didn’t. Or the really seamless and much easier ability to edit within a scene. All are characteristics of what we can do with synthetic media. 

We did a series of threat modeling and solution prioritization workshops globally. In Europe, the US, Brazil, Sub-Sahara Africa, South and Southeast Asia people keep on saying we have to view both types of fakes as a continuum and we have to be looking at solutions across it. And we also need to really think about the wording we use because it may not make that much difference to an ordinary person who is receiving a WhatsApp message whether it is a shallowfake or a deepfake. It matters, whether it’s true or false.

Where do you encounter synthetic media the most at the moment?

Indisputably the greatest range of malicious synthetic media is targeting women. We know that from the research that has been done by organizations like Sensity. We have to remember that synthetic media is a category in the non-malicious, but potentially malicious usages. There is an explosion of apps that enable very simple creations of deepfakes. We are seeing deepfakes starting to emerge on those parody lines, a kind of an appropriation of images. And, at what time does software become readily available to lots of people to do moderately good deepfakes that could be used in satire, which is a positive usage but can also be used in gender-based violence?

Where is the highest impact of deepfakes at the moment?

It is on the individual level. In terms of impact on individual women and their ability to participate in the public sphere, related to the increasing patterns of online and offline harassment that journalists and public figures face.

Four threat areas were identified in our meetings with journalists, civic activists, movement leaders and fact-checkers that they were really concerned about in each region.

  1. The Liars dividend, which is the idea that you can claim something is false when it is actually true which forces people to prove that it is true. This happens particularly in places where there is no strong established media. The ability to just call out everything as false benefits the powerful, not the weak.
  2. There is no media forensics capacity amongst most journalists and certainly no advanced media forensics capacity.
  3. Targeting of journalists and civic leaders using gender-based violence, as well as other types of accusations of corruption or drunkenness.
  4. Emphasis on threats from domestic actors. In South Africa we learned that the government is using facial recognition, harassing movement leaders or activists.

These threats have to be kept in mind with the development of tools for detection. Are they going to be available to a community media outlet in the favelas in Rio facing a whole range of misinformation? Are they going to be available to human rights groups in Cambodia who know the government is against them? We have to understand that they cannot trust a platform like Facebook to be their ally.

Julia Bayer (DW) in conversation with Sam Gregory (WITNESS)

Can synthetic media be used as an opportunity as well?

I come from a creative background. At WITNESS the center of our work is the democratization of video, the ability to film and edit. Clearly these are potential areas that are being explored commercially to create video without requiring so much investment.

I think if we do not have conversations about how we are going to find structured ways to respond to malicious usages, I see positive usage of these technologies being outweighed by the malicious usage. And I think there is a little bit too much of a ”it will all work itself out approach” being described by many of the people in this space.

We need to look closely at what we expect of the people who develop these technologies: Are they making sure that they include a watermark? Do they have a provenance tree that can show the original? Are they thinking about consent from the start?

Although I enjoy playing with apps that use these types of tools, I don’t want to deny that I think 99% of the usage of these are malicious.

We have to recognize that the malicious part of this can be highly damaging to individuals and highly disruptive to the information ecosystem.

Video take

Should we use synthetic media in satire for media literacy? 

We have been running a series of web talks called deepfakery. One of the main questions is: what are the boundaries around satire? Satire is an incredibly powerful weapon of the weak against the powerful. So, for example, in the US we see the circulation of shallowfakes and memes made on sites that say very clearly on the top that this is satire. But of course no one ever sees that original site. They just see the content retweeted by President Trump in which case it looks like it is a real claim.

So satire is playing both ways. I do think the value of satire is to help people understand the existence of this and to push them to sort of responsibly question their reaction to video.

I think the key question in the media literacy discussion is: how do we get people to pause? Not to dismiss everything but to give them the tools to question things. Give them the tools to be able to pause emotionally before they share.

From a technology point of view, what are we still missing to detect synthetic media?

Synthesis of really good synthetic media is still hard. So synthesizing a really good face swap, or a convincing scene is still hard. What is getting easier is the ability to use apps to create something that is impactful but perhaps not believable. I think sometimes people over-assume how easy it is to create a deepfake.

We’re not actually surrounded by convincing deepfakes at this point.

A lot of our work has been thinking about detection and authentication. How do you spot evidence of media manipulation which could be detection of a deepfake or detection of a shallowfake? How to spot that a video has been miscontextualized and that there is an original or an earlier version that has different edits? Then authentication, how do we trace a video over time to see its manipulations?

At the moment the detection of synthetic media is, and this is the nature of the technology, an arms race between the people who will develop the detection tool and those who will use it to test and enhance their new synthesis tool. The results of detection tools are getting better, but they are not at the level that you could do it at scale.

The meta question for us on detection is actually who to make this accessible to. If it is only the BBC, Deutsche Welle, France 24 and New York Times, that leaves out 90% of the world as well as ordinary people who may be targeted by this in an incredibly damaging way.

Video take

Do all journalists need to be trained in using advanced forensic technology?

One of the things we have learned as we have been working on deepfakes is that we shouldn’t exclusively focus on media forensics. I think it is important to build the media forensic skills of journalists, and it is a capacity gap for almost every journalist to do any kind of media forensics with existing content. I do not think we can expect that every journalist will have that skillset. We also need to consider how we invest in e.g. regional hubs of expertise.

The bigger backdrop is that we need to build a stronger set of OSINT skills in journalism. We need to be careful not to turn this purely into a technical question around media forensics at a deep level because it is a complicated and specialist skill set.

We identified a range of areas that need to be addressed to develop tools that plug into journalistic workflows. For example, journalists must not rely on tools easily. They do not need just a confidence number, they need software to explain why something is coming up with a particular result. So, I think we need a constant interchange between journalists and researchers and tool developers and the platforms to say what the tools are that we really need, as this gets more pervasive. And we need tools that potentially provide information to consumers and community leader level activists to help them do the kind of rapid debunking and rapid challenging of the kind of digital wildfire of rumors that journalists frankly often do not get to. Often community leaders are talking about things that circulate very rapidly in a Favela or a Township and journalists never get to them in a timely way. So we need to focus on journalists, but also on community leaders.

What are your three tips for consumers to deal with synthetic media? 

  1. Pause before you share the content.
  2. Consider the intention of why people are trying to encourage you to share it.
  3. To take an emotional pause when consuming media trying to understand the context of it is supported by a range of tools like the SIFT methodology or the Sheep Acronym.

I don’t think it is a good idea to encourage people to think that they can spot deepfakes.

Video take

The clearest and most consistent demand we heard primarily from journalists and fact checkers is to show them if this is a mis-contextualized video so that they can then just clearly say, no this video is from 2010 and not from 2020.

Therefore, reverse video search or finding similar videos is pretty important because that shallowfake problem remains the most predominant.

Many thanks Sam!

*****

More material: Here’s the ‘Ticks or it didn’t happen‘ report that Sam mentioned.

Contact and more: If you are interested to learn more about deep and/or shallowfakes, or have questions on the topic or related issues, please do not hesitate to get in touch with the DIGGER project team (Julia Bayer and / or Ruben Bouwmeester). You may also want to follow the project’s Twitter channel for updates and further information, or visit the DIGGER project website.

Leave a Comment

sing in to post your comment or sign-up if you dont have any account.

WeVerify.eu Privacy Policy

1. Purpose

The purpose of this Privacy Policy is to describe what we collect, use and sometimes share information about you through our online interfaces (e.g., websites and email) owned and controlled by us, including WeVerify and all subdomains (collectively referred to herein as the “Site”).

At WeVerify, we believe that you should have control of your data. Control starts with information. This is why you should know what data we collect from you and how we use it.

This notice and the accompanying policy is to enable you to make the best decisions about the information that you choose to share with us.2

2. Privacy Policy

By accessing and using any of WeVerify site, demonstrators or publicly available services, you expressly and knowingly consent to the information collection and use practices as described in this Privacy Policy.

3. Our Privacy Commitment

Our commitment to your privacy, is based on the following principles which we apply to our use of both your personally identifiable data (“Your Personal Data” or generally “Personal Data”) and to certain anonymous information we collect when you visit our Sites (“Technical Information”, and together with Personal Data, “Your Data”):

  • We will describe Your Data we will collect;
  • We will inform you clearly about our collection and use of Your Data;
  • We will either seek your express informed consent or rely on other legally permissible bases for the use of Your Data – either way, we will inform you of the basis for our use of Your Data;
  • We will give you control over the privacy preferences that apply to Your Data, including the rights to (a) change your mind about our use, (b) have access to change or correct inaccurate aspects of Your Data, and (c) require that we delete all or parts of Your Data (d) request Your Data in a portable format;
  • We will not sell or rent Your Personal Data to others;
  • We endeavor to maximize the protection of Your Data, and provide you with prompt notice in the unlikely event that a data loss incident or breach occurs; and
  • We will endeavor to be completely transparent and open about our data privacy policies and practices.

4. What Information does this Privacy Policy cover?

This Privacy Policy covers information we collect from you through all of our channels, including website, email and others. Some of our website’s functionality can be used without revealing any of Your Data. In order to access certain products, demonstrators or services, you may need to submit, or we may collect information that can be used to identify you.

Your Personal Data can include information such as your name and email address, among other things. You are responsible for ensuring the accuracy of the information you submit to us. Inaccurate information may affect your ability to use the site, download products, any follow-up information you request, and our ability to contact you. For example, your email address should be kept current because that is one of the primary manners in which we communicate with you.

5. How do we collect information?

We collect Your Data in the following ways:

  • You give it to us when you download software or documentation; register for an event such as a webinar; sign up for a newsletter; fill a form on the website or via any other sales or marketing channel;
  • You give it to us by email or phone inquiries or
  • We automatically collect Technical Information when you visit our Sites.

6. What information do we collect?  

When filling any form on the website, we collect Personal Data such as name; phone; email; company name, website and address; job title and category; social media data and nature of the interest.

In addition, we enrich the Personal Data above with Technical Information, related to:

  1. Conversion point (when, where, what campaign, source);
  2. Activity (dates of contact, email opens, link clicks, website visits, etc.);
  3. Opt-in Date to trace your consent;
  4. CRM identifiers;

When processing payments we additionally collect VAT ID, company identification and other information for invoicing and tax purposes.

7. How do we use Your Data collected at our sites?

We will use Your Data to:

  • Provide information, product or a service requested or consented to by you.
  • Comply with relevant contractual obligations with you and other third parties.
  • Improve Site performance and content, including troubleshooting and diagnostics.
  • Improve our engagement and interaction with you.
  • Facilitate your attendance at and participation in our events, communities or blogs.
  • Process a request or payment submitted to us.
  • Comply with legal requests.

8. What are your rights to control Your Data?

You have the right to request that we:

  • provide access to any of Your Personal Data we hold about you;
  • prevent the processing of Your Personal Data for direct marketing purposes;
  • update any of Your Personal Data which is out of date or incorrect;
  • delete Your Personal Data which we are holding about you;
  • restrict the way that we process Your Personal Data;
  • provide Your Personal Data to a third party provider of services; or
  • provide you with a copy of Your Personal Data which we hold about you.

We try to answer every email promptly where possible and provide our response within the time period stated by applicable law. Keep in mind, however, that there will be residual information that will remain within our databases, access logs and other records, which may or may not contain Your Personal Data. Please also note that certain parts of Your Personal Data may be exempt from such requests in certain circumstances, which may include if we need to keep processing Your Personal Data to comply with a legal obligation.

When you email us with a request, we may ask that you provide us with information necessary to confirm your identity.

8. What data do we retain?

We will only retain Your Data stored on our servers in accordance with the legitimate needs of our business and as required or permitted by applicable law. We will not retain any unused Personal Data on our systems longer than necessary for legitimate business purposes.