PhishEye

AI powered phishing prevention that also teaches users how to be vigilant for phishing attempts when browsing.

Comment

Inspiration

Phishing attacks are one of the most common reasons for data breaches across multiple industries. Phishing attacks primarily try to exploit human error. We wanted to explore a possible solution by utilizing breakthrough AI technology to both detect phishing attempts with higher accuracy than the human eye and educate internet users for further prevention.

What it does

PhishEye will leverage existing detection systems and AI learning to catch even the smallest signs of phishing in e-mails. On top of that, the AI would warn the user of what it detected with both visual indicators and explanation of why it flagged that part of the e-mail. Because of how common and dangerous phishing attacks are, we want to implement an educational aspect to inform users of the dangers that may go unnoticed when using their e-mail. That way, they can become more aware of phishing attempts on themselves and others. We hope that with further development and widespread implementation that we can shift the cyber kill chain further towards the point of origin and promote a higher level of phishing-awareness across the internet.

How we are building it

We created an API that extracts the source code of an e-mail for AI analysis. We then plan to train an AI model with both safe and dangerous e-mails so it can decipher whether an e-mail is safe and what signs to look for and flag as dangerous. The training process would be carefully curated to ensure that the AI learns and develops instead of just processing data as well as ensuring that bad habits don't form as, much like humans, bad habits die hard. This AI would then highlight pieces of the e-mail that were flagged as dangerous and give a brief description as to why it flagged the information. The interface would be user-friendly and easily legible providing accessibility for all possible users.

Challenges we have ran into

As a group, we started this endeavor highly inexperienced with AI technology, as is most of the world. We also had little experience with APIs aside from base level knowledge. While we couldn't fully develop the full AI system due to time constraints, we still wanted to challenge ourselves and develop a proof of concept for a solution to a problem we were truly passionate about. By leveraging the power of AI and incorporating it with concepts we were familiar with, we came up with a clear roadmap that gives a sneak peek at the future of phishing prevention.

Accomplishments that we're proud of so far

We successfully created an API that extracts the source code of selected e-mails and returns that data as plaintext to be utilized for training an AI model. We also came up with some of the business aspects of this concept such as a logo and website branding, including getting our own domain to utilize for a website. We are also proud of the effort each of us have put towards PhishEye and how it has turned out in the limited time we have at RevolutionUC.

What we have learned

First and foremost, we learned how powerful AI can be in aiding with the development process and got a sneak peek at how it can be used to develop the future of the world around us. Through our utilization and witnessing others utilize the same technology, we have a better concept of what AI does and can do for the world. We also got a great insight as to how APIs interact with other applications and how, when implemented correctly, they can power some of the most useful and revolutionary systems of today. We also learned how quickly an idea can develop into a reality with the right mindset. Going from limited knowledge of the technologies behind our product, to having a proof of concept with an API demo within the span of 24 hours is something we never thought we could do prior to RevolutionUC. However, we set our minds and our bodies to our concept and it really came out as something we are proud of. Development of new and exciting concepts like PhishEye is extremely fun and gave us confidence and desire to further develop concepts as possible business ventures in the future.

What's next for PhishEye

We plan on fully developing this concept through training of the AI model with phishing examples ranging from common, easy to spot phishing, to indistinguishable from legitimate messages. Hopefully, through careful training and consistent results, we will be able to present the fully-fledged PhishEye AI to major e-mail services such as Gmail and Outlook (Google and Microsoft). We also want to make this design accessible to all internet mail services so everyone can utilize this technology. We want to ensure as many people are able to utilize this revolutionary technology as possible and we will strive to produce and upkeep the best end-product we can. Through PhishEye, the future of internet security looks_ picture perfect.

Built With

Share this project:

Updates