The game is open source and submissions are periodically released to the public for research purposes. This forms the basis for a prompt injection robustness benchmark. The researchers aim to use this data to help build more secure AI systems.
Key takeaways:
- Tensor Trust is a game where players defend their AI-powered bank account and attempt to hack into others' accounts.
- The game is designed to help researchers at UC Berkeley learn more about the vulnerability of AI to a class of attacks called prompt injection.
- Players can increase their account balance by successfully defending or attacking, and rise to the top of the Tensor Trust leaderboard.
- Submissions to Tensor Trust are periodically released to the public, forming the basis for a prompt injection robustness benchmark.