Using browser extensions to uncover disinformation
Trust Tools
© Photo by Jorge Franganillo on Unsplash
Fake information is experiencing a boom, but given the right tools, you can reliably separate the wheat from the chaff.
As the volume of information on the Internet increases, so does the volume of misinformation. It is almost impossible to check all the information you read every day. Even media companies find it difficult to correctly classify and evaluate all the information coming in from the various social media channels.
In the meantime, political groups, fringe actors, and hostile foreign states have made a science out of passing disinformation intentionally to further their political ends. Normal users are finding it increasingly difficult to distinguish genuine news from manipulated news.
The good news is that various developers have identified this shortcoming and are providing extensions for popular web browsers to help distinguish between fake and genuine information. Many of these tools are based on artificial intelligence techniques, but some include support for manual checks or database comparisons for cases where artificial intelligence is not yet up to the task of fully automated checking.
Fake information also affects product reviews, where a company's aim is to sell more of its own items or discredit competitors' products. Browser extensions can scan product reviews, especially on large platforms like Amazon, and alert you to false information.
The currently available browser extensions focus on different areas that serve as the distribution channels for fake news (see the box entitled "Techniques"). Some of the add-ons are exclusively designed for use on the major social networks. Others focus on checking YouTube videos. Others check images published on the Internet, making it easier to expose images that have been altered or misrepresented.
Techniques
Many of the browser extensions for finding fake content use artificial intelligence methods. For example, on text pages, certain phrases, such as "in my opinion" or "I think," already allow conclusions to be drawn about the content's objectivity. These browser extensions also often use databases to compare messages and analytically check their truthfulness.
On Facebook and other social networks, the add-ons can also be used to sort out comments that appear to have been posted by bots. When checking images, some extensions scan the media and check whether an image has already been published elsewhere. If you find the same image on other earlier web pages, and it is not marked as an archive or icon image on the page you are checking, someone might be trying to mislead you. You can also search for other copies of the image, a practice known as a reverse image search, to detect copyright infringement.
Fake videos are far more difficult to detect, and a manual check is usually still required. Services such as CaptainFact [3] offer an online forum for collaborative verification and annotation of videos as well as other web images.
Amnesty International's Citizen Evidence Lab [4] includes an online video verification facility that extracts metadata. If the video is found to be from a far earlier date than the content it purports to describe, you are very likely looking at a fake.
If you really want to cover all the bases, you might need to install several of these browser extensions. It is important to note that some of these extensions are only available for Chrome and its derivatives. Some extensions will also run on Mozilla Firefox, but if you need to access the full range of these services, it is a good idea to keep a Chrome-based browser on hand.
When using the extensions discussed here, you should pay attention to privacy. Some of these extensions require you to log in with a Google account, while others want Twitter and Facebook accounts. Because some of these extension providers do not go into detail about what data they use for what purposes or whether they aggregate data to generated profiles, you might want to avoid such extensions if you're concerned about privacy.
RevEye and TinEye
RevEye [1], a browser extension for Chrome and Firefox, makes it easier to check images with the help of several search engines. After installation, activate the image search by right-clicking on the image you wish to check and select the Reverse image search entry from the context menu that opens. You then need to click on one of the four image search engines: Google, Bing, Yandex, or TinEye.
TinEye [2] is an engine that specializes in image searches and does not store any data. The fact that TinEye doesn't store user data makes it especially suitable for security-conscious users. (TinEye is also available as an extension for Firefox and Chrome.)
Click on the All search engines option in TinEye to open four tabs in the browser that display the results from all four search engines. In testing, TinEye's results were clearly superior to those of the other engines. For example, Bing displayed a hodgepodge of similar images in the test, but hardly any exact matches. In addition, Bing also lacked most of the relevant data for the individual images. The results for Yandex and Google were also less meaningful than the TinEye results. TinEye, on the other hand, listed the virtually identical images it found, including the metadata, and also provided a comparison of the matches with the original (Figure 1).
To compare the matches with the subject of the search, move the mouse pointer to one of the listed images and then click on the Compare link. The image will then appear in a separate window. To view the images in comparison, click on Image match or Your image.
TinEye also displays the publication date to the right of each image in a list, as well as the origin URL and the image size. This summary of statistics in list form makes it easy to determine whether an image has actually been created recently – or whether the image has been manipulated by subsequent retouching. TinEye also lets you track down copyright infringements on images.
The Factual
The Factual [5] is available as a browser add-on for Chrome and its derivatives and can be displayed using the toolbar after installation. The Factual portal, which belongs to the California-based CivikOwl organization and provides the browser add-on of the same name, also relies on a mixture of artificial intelligence and manual checking for its text analysis; it also references databases.
The Factual bills itself as the "world's largest news ratings engine." The project claims to have analyzed 10 million news stories from 50,000 journalists and 2,000 news sources to develop a grading system that evaluates a news story based on four factors:
- Diversity and extent of sources
- Author's tone (neutral vs. opinionated language)
- Author's expertise on topic
- Site's historical reputation
The browser extension opens a small window when launched. Colored tiles appear after completing the analysis, with the tiles providing visual information on the status of the article that was checked. The four tiles available in the basic setting evaluate the general quality of the investigated medium, the expertise of the article's author, the quality of the sources used, and the objectivity of the article. Using an overall quota, the extension also presents the quality of all factors as an absolute percentage value and identifies the political orientation of the medium (Figure 2).
The Show details link displayed at the bottom of the window lets you view more detailed information about the reviewed article. You will then also see the number of sources and links mentioned. The color spectrum of the individual evaluation criteria, ranging from green to red, helps you see at a glance how to rate the article (Figure 3).
TrustedNews
TrustedNews [6], another browser extension, uses artificial intelligence to analyze media articles on English-language websites. The tool features an Objectivity category that checks an article's objectivity and highlights the more objective sentences in yellow. In addition, it gives a value for a news story's objectivity on a scale of one to five (Figure 4). The lower this value, the more subjective the report. TrustedNews also determines an additional value for the medium based on multiple criteria. For example, users can press a radio button to say whether they feel an article is objective or subjective in terms of its content.
Buy this article as PDF
(incl. VAT)
Buy Linux Magazine
US / Canada
UK / Australia