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 4 of 15
Current
SeverityCVECVSS
HIGHCVE-2021-412287.8
HIGHCVE-2022-217308.1
MEDIUMCVE-2022-235636.3
HIGHCVE-2022-235738.8
MEDIUMCVE-2022-291935.5
MEDIUMCVE-2022-292075.5
MEDIUMCVE-2022-292115.5
HIGHCVE-2022-359647.5
CRITICALCVE-2022-418809.1
HIGHCVE-2022-418977.5
CRITICALCVE-2022-419109.1
HIGHCVE-2023-256587.5
HIGHCVE-2023-256747.5
HIGHCVE-2023-275067.8
MEDIUMCVE-2023-307676.7
HIGHCVE-2021-41187.8
CRITICALCVE-2022-08459.8
CRITICALCVE-2024-480639.8
MEDIUMCVE-2025-29985.3
MEDIUMCVE-2025-29995.3

Page 4 of 15