AI Component

Training Data

Training data is both the model's most valuable input and its most underprotected one. Three problem classes dominate. First, poisoning: an attacker who can influence a public dataset, a web crawl, or a fine-tuning corpus can plant backdoors or biases that survive into the deployed model — BadNets-style attacks on image classifiers, trigger-phrase attacks on LLMs, and reward-hacking on RLHF datasets. Second, memorization and leakage: models can regurgitate verbatim training data, exposing PII and copyrighted content; this has driven the active New York Times v. OpenAI litigation and is a recurring GDPR concern. Third, provenance: when training data origins are unclear, downstream users inherit legal and security risk they can't assess. EU AI Act Article 10 (Data Governance) and ISO 42001 Annex A treat training-data quality as a controlled asset. Defenses: data lineage tracking, deduplication, PII scrubbing before training, and adversarial training against known trigger families.

287
Total CVEs
15
Pages
Page 7 of 15
Current
SeverityCVECVSS
HIGHCVE-2025-465677.8
HIGHCVE-2025-617848.1
MEDIUMCVE-2024-52064.7
MEDIUMCVE-2024-554596.5
CRITICALCVE-2025-120609.8
UNKNOWNCVE-2025-12638-
HIGHCVE-2024-435988.1
MEDIUMGHSA-5cxw-w2xg-2m8h-
LOWGHSA-83pf-v6qq-pwmr-
HIGHCVE-2026-17777.2
MEDIUMGHSA-m7j5-r2p5-c39r-
HIGHCVE-2026-22033-
HIGHCVE-2026-22612-
HIGHGHSA-9726-w42j-3qjr-
CRITICALCVE-2025-332449.0
CRITICALCVE-2025-34351-
MEDIUMCVE-2025-89175.8
CRITICALCVE-2023-480229.8
HIGHCVE-2025-587578.8
HIGHCVE-2025-587568.8

Page 7 of 15