Generative AI - Hangman Game

Top 10 AI Vulnerabilities

• Prompt Injection – Manipulating AI through deceptive inputs.
• Insecure Output Handling – Accepting AI output without validation.
• Training Data Poisoning – Injecting malicious data into training sets.
• Model Denial of Service – Overloading AI systems to degrade performance.
• Supply Chain Vulnerabilities – Risks from third-party models or datasets.
• Sensitive Information Disclosure – Accidental leakage of private data.
• Insecure Plugin Design – Plugins allowing unauthorized access or code execution.
• Excessive Agency – AI with too much autonomy causing unintended actions.
• Overreliance – Blind trust in AI leading to misinformation or poor decisions.
• Model Theft – Unauthorized access to proprietary AI models.