Cyberdojo
We equip you for a successful cloud security journey
RESEARCH
Why the focus on GenAI?
-
The Need for Security in GenAI Applications: Generative AI (GenAI) applications are revolutionizing industries by automating tasks, generating content, and enhancing decision-making processes. However, these advancements come with significant security challenges. GenAI systems are vulnerable to various threats, including data poisoning, adversarial attacks, and misuse of generated content. Ensuring the security of GenAI applications is critical to prevent malicious exploitation, protect sensitive information, and maintain the integrity and trustworthiness of AI-generated outputs.
-
Leveraging GenAI for Enhanced Security: The rise of GenAI is also transforming how we approach cybersecurity. GenAI tools can enhance threat detection, automate response strategies, and predict potential vulnerabilities with unprecedented accuracy. By integrating GenAI into our security frameworks, we can stay ahead of emerging threats and continuously adapt to the evolving digital landscape.
Project hackerBot
hackerBot is an AI-driven cybersecurity tool based on OpenAI’s models, designed to perform various cybersecurity tasks. It can be run in a Docker container or installed locally. The tool is equipped with skills such as AWS CLI, port scanning using nmap, Netcat, and reading AWS logs using LangChain Agent. It allows users to execute custom commands with or without AI assistance, offering flexibility and control.
Project Aviata-chatbot
Aviata-chatbot is a purposefully designed vulnerable Generative AI (GenAI) application created to investigate and analyze potential security issues that could arise in GenAI systems. By intentionally incorporating known vulnerabilities, Aviata-chatbot allows us to study a wide range of security threats, such as data breaches, adversarial attacks, unauthorized access, and the manipulation of AI outputs. This research is crucial for understanding the specific risks associated with GenAI applications and for developing effective strategies to mitigate these risks and enhance the overall security of AI-driven technologies.
Project Cloudwatch-bot
Cloudwatch-bot is a proof-of-concept project that demonstrates how AI can be utilized to interface with security solutions. The project has a user interface that is built using HTML and JavaScript and is hosted on a public S3 bucket. The UI communicates with a backend system that includes an API Gateway and a Lambda function, which is written in Python and has permission to access OpenAI and CloudWatch. When a user makes a request, the API Gateway triggers the Lambda function, which translates the request using OpenAI into a CloudWatch query that searches for relevant information in CloudWatch logs. Use it LIVE here.
AWS CloudWatch-bot Sample Code