Data poisoning
Data poisoning is holding steady in AI research: 0.1% to 0.1% of new AI papers, 0.0 points. As of Oct 10, 2026.
- Change in share
- 0.0 pts
- Share, last 3 months
- 0.1%
- Papers, last 3 months
- 29
- Ready for products
- 2.3 / 5
More in Adversarial robustness and AI security
Retrieval augmented generation security0.1% → 0.1%+0.0 ptsAI generated text detection0.1% → 0.2%+0.0 ptsJailbreak defense0.1% → 0.2%+0.0 ptsVisual prompt injection0.0% → 0.0%0.0 ptsSoftware supply chain security0.0% → 0.0%0.0 ptsUniversal adversarial perturbation0.0% → 0.0%0.0 ptsDeepfake detection0.1% → 0.1%0.0 ptsTransferable adversarial attack0.0% → 0.1%0.0 ptsMembership inference attack0.1% → 0.1%0.0 ptsPrivacy auditing0.1% → 0.1%0.0 pts
Data poisoning: quick answers
No. Its share of new AI papers went from 0.1% to 0.1%, with 29 papers in the last 3 months.