Prompt injection or adversarial prompting can override the attention of Large Language Models to previous instructions and force them to act on the current prompt, an issue that has affected GPT-3 (Branch et al. 2022).

Authors

• Person: Not available Organization: arXiv
  Building Trustworthy NeuroSymbolic AI Systems - arXiv

Sources

• Building Trustworthy NeuroSymbolic AI Systems - arXiv arxiv.org arXiv via serper

Referenced by nodes (2)

• Large Language Models concept
• GPT-3 concept