WhiteRabbitNeo, a new AI model series designed for cybersecurity applications, has launched its 33B model in beta. This release serves as a public preview to showcase the model’s capabilities and to evaluate its societal impact. The model is accessible online, and interested parties can also join the community on Discord.
The model comes with a strict license that prohibits its use in various scenarios, including any activity that violates laws, infringes on third-party rights, exploits minors, spreads false information, or discriminates against individuals or groups based on protected characteristics. Notably, the model cannot be used for military purposes, fully automated decision-making with legal consequences, or to exploit vulnerabilities of specific groups.
WhiteRabbitNeo is implemented using PyTorch and the transformers library, and it can be used to generate text based on given instructions. The sample code provided demonstrates how to load the model, tokenize input, and generate text, which can be used in various cybersecurity tasks.
A sample conversation illustrates how the model can guide users through the process of attacking a Wi-Fi network, covering steps from network scanning to post-exploitation. It emphasizes the importance of ethical hacking, stating that such activities should only be conducted on networks where the user has explicit permission to avoid legal and ethical violations.
Read more…