Semantic vs. Token-Based LLM Injections
blog.sentry.security | blog | #ai-security | #prompt-injection | #llm | #owasp | #tokenization | #adversarial-ml
Summary
Contrasting two LLM injection classes: semantic attacks use natural language to override instructions; token-based attacks exploit tokenizer quirks like reserved delimiters and adversarial suffixes.
- Published
- Collected
--- id: 7448 title: "Semantic vs. Token-Based LLM Injections" source_domain: blog.sentry.security source_url: "https://blog.sentry.security/semantic-vs-token-based-llm-injections/" published_at: "2026-05-12T09:22:53Z" tags: ["ai-security", "prompt-injection", "llm", "owasp", "tokenization", "adversarial-ml"] phying_url: "https://phying.com/item/7448" lang: en --- # Semantic vs. Token-Based LLM Injections ## Summary Contrasting two LLM injection classes: semantic attacks use natural language to override instructions; token-based attacks exploit tokenizer quirks like reserved delimiters and adversarial suffixes. ## Intelligence Metadata - **Sources:** blog.sentry.security - **Published:** 2026-05-12T09:22:53Z - **Original:** https://blog.sentry.security/semantic-vs-token-based-llm-injections/ - **Phying News:** https://phying.com/item/7448