PNAS Study: Social Algorithms Prioritize Content Clashing with Your Values
msbernst · x · 2026-08-21
A new PNAS study reveals a misalignment in social media recommendation algorithms, finding that X's "For You" feed often prioritizes content that conflicts with users' core values, even from accounts they follow.
Key Findings:
- Algorithms rely heavily on engagement, particularly comments, to predict user preferences.
- Users are more likely to comment on content they disagree with (criticism/anger) than content they agree with.
- Algorithms misinterpret this negative engagement as a positive signal of interest, amplifying conflicting content.
- This effect is more pronounced for Democratic users but present in both groups.
Researchers compared user values measured via psychological scales against the values expressed in their feeds, confirming this widespread "value misalignment."
More from Safety
- Vector Institute partners with Town of Caledon in its first municipal AI adoption deal — VectorInst · 2026-10-16
- 'Most dangerous technology in the world' put in toddler sandboxes, and we're surprised when it escapes — KevinNaughtonJr · 2026-10-16
- NYC Comptroller urges Anthropic to pause IPO until AI safety and governance concerns are addressed — ShakeelHashim · 2026-10-16
- Dwarkesh: Anthropic insider expects AI to match humans at AI research within 9 months — ShakeelHashim · 2026-10-16
- Anthropic and Gray Swan launch $300,000 indirect prompt injection red-teaming contest — bcherny · 2026-10-16
- Experts: AI's nearer bioweapons risk is amplifying capable actors, not enabling amateurs — HaydnBelfield · 2026-10-16