Introduction

We live in a world dominated by technology, where social media has taken up a considerable chunk of our time and minds. An average user spent 2 hours and 24 minutes daily on social media in 2020 (We are Social) and by 2023 there will be 5.3 billion total internet users consuming media (cisco). Audiovisual content on the internet has also surged exponentially since digital marketing and communications are replacing the traditional means. It is estimated that by 2022, 82% of all the consumer web traffic will be video (cisco). Generation of synthetic audiovisual content via artificial intelligence is another trend on rise and this generated media is popularly referred to as deepfakes.

The listed staggering figures alludes to the progression and accessibility of information to masses. Now let’s consider for a moment what if most of the video content that we will come across becomes synthetic and is generated with such sophistication that we can no longer differentiate between real and synthetic content. This scenario is not far fetched and the future of online media is majorly synthetic. There are serious concerns that arise with this technological advancement, such as fake news, the spread of disinformation via deepfakes to name a few. Viewers retain 95% of a message through audiovisual content (insivia). What we see and hear build our perception and thus it becomes imperative to know if the content that is being consumed is genuine or not. 

Deepfakes can depict someone saying or doing something, which they may not have said or done in real life. Deepfakes hold enormous power to spread misinformation affecting companies and individuals involved with social media, political parties, celebrities, PR agencies, media houses, law firms, cybersecurity firms, and government agencies. During the recent US Presidential elections, it became a significant variable and corporations like Facebook and Twitter came under the government’s purview, where political parties accused the other of promoting such deep fakes on social media to leverage votes. The FBI has recently warned private business to be cautious of potential harm, which may come through deepfakes.

To counter this problem, Kroop AI has the perfect solution. Kroop AI offers a state-of-the-art explainable deepfakes detector for audio and video data. It is an Indian DeepTech AI startup at the forefront of fighting disinformation using AI. Kroop AI's powerful multimodal deep learning based platform detects and analyzes deepfakes in detail across different platforms and information mediums. The solution also provides detection of subtle manipulations and a rich analytics interface to the user. 

Where it all began

After over a decade in academia research, Jyoti was contemplating how her research could have a broader impact. Jyoti and Abhinav came across a high quality deepfake of a state premiere and realised the potential havoc which misinformation could cause. Abhinav introduced Jyoti to Sarthak and Milan, two extremely talented deep learning engineers, whom he knew from IIT Ropar. Together the team narrowed down the problem of deepfakes detection as a service. 

People have been using deepfakes with malicious intent to spread disinformation and there are very few companies/solutions that help detect these deep fakes. That’s when they realized that this is a gap to be addressed with a lot of potential to grow across different sectors.

Market Opportunity 

In India, this sector is interwoven with cybersecurity. The Indian Antivirus Software market is projected to reach INR 14,782 crores by 2024. Deepfake technology falls under this market and is emerging as something that will be crucial to society in the coming days. A study estimates that the market in India for cyber security products will grow at a higher rate than that for services, as organisations invest more in products powered by specialized technologies.

Kroop AI has four different products to cater to different stakeholders:

1. API integration that helps any video streaming platform to check for deep fakes and authenticity of the video by AI.
2. A web app that gives a detailed analysis of a video or an audio for its authenticity. It’s a low cost developer free solution helpful to media houses and law firms.
3. A web scraper tool, that crawls through the web, based on the information like keywords and geographic location to search for such videos and audios before it gets viral. It will be useful for all the political parties and government agencies to take down such content before it spreads.
4. Lastly, a mobile app to give offline analysis on the go. All these four different products will cater to a huge market and help most of the stakeholders involved on a very large scale.

100X.VC Thesis 

Considering how rapidly deep-fakes are going to increase in the current market, it is a real problem that needs to be addressed before the situation gets out of control. Kroop AI solves this problem brilliantly by catering to all the stakeholders with their four products. Using the deep AI technology and recognizing the patterns, the company keeps on improving their algorithm to catch such deep fakes and the subtle inconsistencies. 

We believe that the company solves a very critical problem with great IP of the founders who hold the State of the Art in Deep Learning. Taking into account how deep-fakes leverage powerful techniques from machine learning and artificial intelligence to manipulate or generate visual and audio content; to counter such deep tech, the right knowledge and tech is required. The founders with their academic background and entrepreneurial bent, are the  right ones to create a Deep Tech IP driven company solving a problem of the future.

Conclusion

With the first mover’s advantage in Asia pacific, Kroop AI can acquire a majority of the market share. We believe the founding team is well versed with the technology and is full of potential. We are excited for their launch and journey ahead and wish them luck!