Every day millions of selfies are posted and shared, without a thought given to the risks they may pose. But what few realise is that advances in technology have turned every camera and selfie taken with it into a data mine ripe for hackers to excavate. Today, Ubisoft and AKQA launch the Watch Dogs 2: Selfie Reveal to help expose those hidden dangers.
The Selfie Reveal is a first-of-its-kind digital experience that uses cutting edge technology to uncover the risks of seemingly innocent selfies by detecting the image content of your photo through multiple image analysis APIs, and then demonstrating how hackers, companies, or even the government could use the data to harm you. This platform can recognise your expressions, body language, location, and other objects in the photo that a proprietary algorithm uses to draw hypothetical conclusions that are displayed back to you.
In Watch Dogs 2, you’ll explore these relationships between entities and the tech revolution as Bay Area hacker, Marcus Holloway, and while the game is animated, the threats are real.
Find out what’s being revealed about you with Selfie Reveal here. WatchDogs 2 launches on November 15th.
How the Experience Works:
What is it?
A first-of-its-kind digital experience that uses cutting edge computer-learning technology to expose the risks of seemingly innocent selfies.
The Selfie Reveal engine leverages a powerful machine learning platform to detect the image content of your selfie, then demonstrates how corporations, governments, and hackers could use this data against you. This platform can recognize your expressions, body language, facial details, location, emotional state and other physical objects in the photo that a proprietary algorithm uses to draw hypothetical conclusions that are displayed to the user.
How does the site work?
It runs a selfie (or any photo) through artificial intelligence processers that can recognize certain objects and information contained in photos. The site uses this information recognized by running it through custom algorithms and comparing it against publicly available databases to draw possible conclusions that are displayed to the user.
What personal information do we have access to?
Just the photo uploaded by a user, any meta data associated with that photo and an IP address. There is no connection to any social network or public profile. The results returned are solely based on the information contained in the photo.
How long will this information be stored?
Users can share each result module directly to Facebook and Twitter or download to their mobile device or desktop. We temporarily store information for no longer than 24 hours, then all information and data will be wiped from our servers.
How accurate is the site?
The site will only make an informed guess about what’s depicted in the photograph from what the AI applications recognize and any meta data that may the photo may contain (such as capture location IP address, etc.). All this information exists in almost every single digital photo. The level of confidence in the accuracy of these AI applications is unpredictable and solely dependent on the quality of the uploaded photo and the machine learning technology powering those API’s, which are always evolving.
How long will a “report” be viewable to a user?
Until they leave the page or unless they share or download a module.
Does the site actually share any information with any 3rd parties like Big Pharma or Financial Institutions?
No, the site simply for entertainment only and does not connect to any 3rd parties besides the developer of the AI and UbiSoft. The only info we’re sharing with any third parties is the user’s photo and their IP address.
Will the site connect directly to any social network?
Only the API for sharing on Facebook and Twitter which entails no personal user data.
Where will we store the “shared reports” and who will have access to it?
The image of the report module the user decides to share is stored on the server and anyone with the URL to that image (with a random filename) can view it.
Source: AKQA
You must be logged in to post a comment Login