Search papers, labs, and topics across Lattice.
1
0
2
Current safety mechanisms for LLMs are critically flawed, as a genetic algorithm can effectively evolve adversarial prompts to bypass them entirely.
