llm-security-spans-training-and-inference-surfaces

IN derived (depth 1)

Created 2026-06-21T10:10:04+00:00 · Reviewed 2026-06-21T14:41:08+00:00

LLM security threats operate across both training time (deliberate data poisoning via grooming) and inference time (prompt injection), requiring defense at multiple pipeline stages rather than a single chokepoint.

Summary

The key takeaway is that securing a large language model is not a one-gate problem: an adversary can bias what the model learns before it ships by flooding public data with targeted content, and then manipulate it again after deployment through crafted user inputs. That means a system needs layered defenses spanning the entire pipeline rather than relying on a single checkpoint to catch every threat.

Justifications

SL — Two independent attack vectors targeting different pipeline stages establish a dual-surface threat model

Antecedents (all must be IN):

IN prompt-injection-primary-security-concern — Prompt injection is the primary security concern for deployed LLM applications
IN llm-grooming-deliberate-training-data-poisoning — LLM grooming is the deliberate mass-publishing of web content to bias LLM training data and outputs, a term coined by the American Sunlight Project in 2025 (e.g., the Pravda network).

Dependents

These beliefs depend on this one:

IN llm-security-requires-defense-across-three-independent-surfaces — LLM security threats operate across three independent attack surfaces requiring distinct defenses: training data poisoning (deliberate grooming of web content), architectural prompt sensitivity (40%+ accuracy shifts from formatting, instruction-input confusion), and inference-time injection — and the architectural vulnerabilities are fundamental, not solvable by engineering or scale.