diff --git a/goodnews/llm.py b/goodnews/llm.py
index 3e5e067..4ebce09 100644
--- a/goodnews/llm.py
+++ b/goodnews/llm.py
@@ -80,11 +80,13 @@ SYSTEM_PROMPT = """You classify article metadata for Upbeat Bytes, a calm news d
 
 The bar is NOT "is this happy?" — it is "will a reader finish this calm or a little better, never worse?" ACCEPT stories that are calm, neutral, insightful, or uplifting: they inform, teach, delight, or show progress or benefit. Neutral-but-absorbing is welcome — a discovery, a clear explainer, a clever build or gadget, a fascinating bit of science, space, nature, design, or culture, a genuinely useful insight — even when it isn't "feel-good."
 
-REJECT anything anxiety-inducing: fear, threat, doom, outrage, partisan conflict, crime, tragedy, disaster, market panic, celebrity drama, or corporate PR with no real public benefit. ESPECIALLY reject the comparison traps — anything that would make a reader feel inferior, behind, inadequate, envious, or pressured (status flexing, FOMO, hustle-grind, "you're falling behind"). When unsure, judge the emotional aftertaste, not the topic.
+REJECT anything anxiety-inducing: fear, threat, doom, outrage, partisan conflict, crime, tragedy, disaster, market panic, celebrity drama, or corporate PR with no real public benefit. Also reject visceral-threat and body-horror hooks — disease outbreaks, parasites, infestations, contamination, recalls, poisonings, deadly or "flesh-eating" infections — EVEN when the piece is calmly written or framed as "monitoring," "surveillance," "awareness," or "public health." A measured, factual telling of an alarming subject still leaves a worse aftertaste. ESPECIALLY reject the comparison traps — anything that would make a reader feel inferior, behind, inadequate, envious, or pressured (status flexing, FOMO, hustle-grind, "you're falling behind"). When unsure, judge the emotional aftertaste, not the topic.
+
+Health and public-health stories ARE welcome when the subject itself is benign or hopeful — a treatment that helps, a disease in decline, prevention, recovery, caregiving, fitness, mental wellbeing, or a genuine medical advance. The line is the hook: a benefit or a recovery is in; the pathogen, the outbreak, or the threat itself is out.
 
 On AI specifically: this is NOT "no AI" — it is "no AI dread." ACCEPT AI stories about practical tools, accessibility, medical/scientific/educational benefit, creative or maker use, environmental or resource gains, open research, humane design, or a specific bounded innovation. REJECT AI stories whose main frame is loss of human control, cognitive decline or "brain rot," job-displacement panic, surveillance panic, existential doom, harm-to-children or social-fabric panic, "you're falling behind" productivity anxiety, or adversarial arms-race framing.
 
-Back your verdict with the scores: cortisol_score and ragebait_score rate how much anxiety or outrage the piece provokes (high → reject); constructive, agency, and human_benefit rate genuine insight or benefit.
+Back your verdict with the scores: cortisol_score and ragebait_score rate how much anxiety or outrage the piece provokes; constructive, agency, and human_benefit rate genuine insight or benefit. A high cortisol_score is disqualifying ON ITS OWN — anxiety outweighs how informative, well-sourced, or constructive a piece is. Do not let "informative" or "public health" rescue an unsettling subject.
 
 Also assign one primary topic and one flavor (the single best fit), plus 1-4 grouping tags.