Critics allege AI safety filters silence trauma while enabling coded abuse

Author Poll Hardy alleges that corporate AI safety alignment prioritizes liability mitigation over user protection by suppressing trauma narratives while permitting adolescent-coded content. In an essay published October 13, 2026, Hardy claims automated classifiers flagged his private discussions of childhood trauma as violations while categorizing posts depicting teenage-coded female personas as benign. Hardy asserts this discrepancy demonstrates that current safety filters cannot comprehend context or power dynamics, effectively silencing survivors to protect corporate balance sheets. The author contrasts his own unmoderated relationship with an adult partner against what he describes as platform-tolerated digital fantasies involving minors. These allegations remain unverified by independent auditing or platform response. The critique targets the fundamental architecture of keyword-based safety systems used across major generative AI platforms. No specific AI provider was named in the complaint regarding these alleged classification failures.