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 3 of 15
Current
SeverityCVECVSS
MEDIUMCVE-2021-376745.5
HIGHCVE-2021-376797.8
MEDIUMCVE-2021-376906.6
MEDIUMCVE-2021-411985.5
MEDIUMCVE-2021-412005.5
HIGHCVE-2021-412107.1
HIGHCVE-2021-412037.8
HIGHCVE-2021-412057.1
HIGHCVE-2021-412117.1
HIGHCVE-2021-412127.1
HIGHCVE-2021-412197.8
HIGHCVE-2021-412237.1
HIGHCVE-2021-412247.1
HIGHCVE-2021-412267.1
MEDIUMCVE-2021-412025.5
HIGHCVE-2021-412087.8
MEDIUMCVE-2021-412185.5
HIGHCVE-2021-412207.8
HIGHCVE-2021-412217.8
HIGHCVE-2021-412257.8

Page 3 of 15