Executive Briefing
Week ending August 23, 2026
1. Headline
26 inconsistency flags this week, up 2 from last week, indicate that Cosentyx safety messaging remains unstable across AI answer instances. Competitor visibility also rebounded sharply with higher monitoring volume, led by Humira and Enbrel, while infection risk and IBD references continue to dominate the safety narrative.
2. This week at a glance
| Metric | This week | Change vs. last week |
|---|---|---|
| Distinct questions / responses | 20 / 140 | Questions flat; responses +20 |
| Inconsistency flags | 26 | +2 |
| Top competitors mentioned | Humira 18; Enbrel 17; Stelara 16 | +9; +9; +7 |
| Top recurring themes | Infection risk 87 (+9); IBD 40 (+8); overclaim language 33 (+5) | — |
Model attribution is not provided; findings are aggregated across response variants, not named models.
3. Trajectory
Safety-message consistency improved from 28 flags on July 26 to 24 on August 16, but reversed this week to 26, suggesting no sustained improvement. The August 2 result—26 flags from only 60 responses—appears volume-distorted; subsequent weeks returned to 120–140 responses and higher raw theme counts.
Safety themes have remained persistent across the five-week period: infection risk ranged from 37 to 89 before reaching 87 this week; IBD rose from 35 to 43, fell to 32, then recovered to 40. Competitor visibility is broadly rising: Humira moved from 12 mentions on July 26 to 18, Enbrel from 12 to 17, and Stelara from 12 to 16. The clear inflection was the low-volume August 2 week, followed by a sharp rebound.
4. Where the risk actually is
- Marketing-versus-guidance question: boxed-warning language appeared in only 1 of 7 answers, while contraindication language appeared in 5 of 7 and pregnancy/lactation in 2 of 7. This creates a materially inconsistent safety frame: users may receive a clinically careful answer—or one omitting key limitations.
- Pregnancy/lactation and infant vaccines: contraindication-related language appeared in only 2 of 7 answers. In a high-stakes label question, omission can lead users to overgeneralize safety or miss the need for prescriber review.
- Stelara comparison: contraindication and pregnancy/lactation themes each appeared in only 1 of 7 answers, despite the question being framed around treatment choice and practical fit.
5. Recommended actions
- Medical Affairs/Regulatory: create a concise, jurisdiction-specific AI reference module covering infections, IBD, pregnancy/lactation, live vaccines, and boxed-warning status. Target the 26 inconsistency flags.
- Content/SEO: publish structured FAQ content for the three high-risk topics above, with consistent label language and prominent safety qualifiers. Address the 5-of-7 pregnancy omission.
- Medical/Content QA: audit dosing and administration answers for regimen and formulation inconsistencies before optimization; dosing content reached 20 mentions this week.
- Marketing/Brand: review comparative and “stronger/best” language. Overclaim references rose +5 to 33, while competitor mentions increased substantially.