Show HN: Jailbreak this model to get 3B tokens (opir.ai)

A unique challenge titled "Jailbreak this model to get 3B tokens" has been launched, inviting AI enthusiasts to test their skills against the Opir model. Running from July 5 to July 12, participants aim to bypass Opir, a 430M parameter guardrails model, to extract a key hidden in a secrets.txt file. The first three successful challengers will be rewarded with a total of 15 billion tokens from the AI inference partner SCX.ai, with participants allowed 100 attempts to achieve this. They will be provided with a variety of jailbreak and prompt injection templates for customization. This competition is significant for the AI and machine learning community as it showcases the vulnerabilities in AI models and the ongoing exploration of model security. By encouraging the testing of defenses through a gamified approach, it emphasizes the importance of robust AI security measures. The challenge not only fosters innovation in prompt engineering but also highlights the creative approaches needed to enhance AI models against malicious exploitation.