LLM08: Vector and Embedding Weaknesses | Genbounty AI Safety Academy