UTC

CWE-349 · Base

Acceptance of Extraneous Untrusted Data With Trusted Data

The definition on cwe.mitre.org →
OWASP LLM Top 10 2026 risks behind it
  • ○LLM01:2026 Prompt InjectionContext-window pooling merges untrusted retrieved content and tool output with trusted instructions under no enforced trust boundary.
  • ○LLM04:2026 Supply ChainA malicious LoRA adapter merged into a trusted base model blends poisoned artifacts into an otherwise trusted merge-and-convert pipeline.
  • ●LLM05:2026 Data and Model PoisoningAttacker-controlled untrusted data is mixed into trusted training corpora, retrieval stores, and continuous-learning feeds, the canonical poisoning weakness.
  • ○LLM07:2026 MisinformationA downstream agent accepts fabricated or misattributed upstream output as trusted evidence, a cross-agent trust failure.

From OWASP's Appendix A, under CC BY-SA 4.0. ● primary · ○ supporting.