# Fine-tuned LLMs

Model weights are adapted using task or domain training examples.

Classification definition is a project adaptation of the DASF deployment model descriptions. Source applicability is column J of [AI Lifecycle Risks](https://docs.google.com/spreadsheets/d/1lD5nKC1QNoUTDXdMp31NA9eKcUYGjhtC34MCfQpqpGo/edit#gid=307775890), snapshot 2026-10-22.

Combine patterns using **union (OR)**. An agent that uses retrieval and a third-party model should include agentic-ai, rag and external-model; include foundation-llm if the model is a foundation model. Do not assume “pre-trained” means merely calling an already trained foundation model: this taxonomy distinguishes pretraining from consumption.

50 source-applicable risks:

- [DASF-R-1.1](https://mathborgess.github.io/ai-security-wiki/wiki/risks/1.1.md) — Insufficient access controls
- [DASF-R-1.2](https://mathborgess.github.io/ai-security-wiki/wiki/risks/1.2.md) — Missing data classification
- [DASF-R-1.3](https://mathborgess.github.io/ai-security-wiki/wiki/risks/1.3.md) — Poor data quality
- [DASF-R-1.4](https://mathborgess.github.io/ai-security-wiki/wiki/risks/1.4.md) — Ineffective storage and encryption
- [DASF-R-1.5](https://mathborgess.github.io/ai-security-wiki/wiki/risks/1.5.md) — Lack of data versioning
- [DASF-R-1.6](https://mathborgess.github.io/ai-security-wiki/wiki/risks/1.6.md) — Insufficient data lineage
- [DASF-R-1.7](https://mathborgess.github.io/ai-security-wiki/wiki/risks/1.7.md) — Lack of data trustworthiness
- [DASF-R-1.8](https://mathborgess.github.io/ai-security-wiki/wiki/risks/1.8.md) — Legality of data
- [DASF-R-1.9](https://mathborgess.github.io/ai-security-wiki/wiki/risks/1.9.md) — Stale data
- [DASF-R-1.10](https://mathborgess.github.io/ai-security-wiki/wiki/risks/1.10.md) — Lack of data access logs
- [DASF-R-1.11](https://mathborgess.github.io/ai-security-wiki/wiki/risks/1.11.md) — Compromised 3rd-party datasets
- [DASF-R-3.1](https://mathborgess.github.io/ai-security-wiki/wiki/risks/3.1.md) — Data poisoning
- [DASF-R-3.2](https://mathborgess.github.io/ai-security-wiki/wiki/risks/3.2.md) — Ineffective storage and encryption
- [DASF-R-4.1](https://mathborgess.github.io/ai-security-wiki/wiki/risks/4.1.md) — Lack of traceability and transparency of model assets
- [DASF-R-4.2](https://mathborgess.github.io/ai-security-wiki/wiki/risks/4.2.md) — Lack of end-to-end ML lifecycle
- [DASF-R-5.2](https://mathborgess.github.io/ai-security-wiki/wiki/risks/5.2.md) — Model drift
- [DASF-R-5.3](https://mathborgess.github.io/ai-security-wiki/wiki/risks/5.3.md) — Hyperparameters stealing
- [DASF-R-6.1](https://mathborgess.github.io/ai-security-wiki/wiki/risks/6.1.md) — Evaluation data poisoning
- [DASF-R-6.2](https://mathborgess.github.io/ai-security-wiki/wiki/risks/6.2.md) — Insufficient evaluation data
- [DASF-R-6.3](https://mathborgess.github.io/ai-security-wiki/wiki/risks/6.3.md) — Lack of Interpretability and Explainability
- [DASF-R-7.1](https://mathborgess.github.io/ai-security-wiki/wiki/risks/7.1.md) — Backdoor machine learning / Trojaned model
- [DASF-R-7.2](https://mathborgess.github.io/ai-security-wiki/wiki/risks/7.2.md) — Model assets leak
- [DASF-R-7.3](https://mathborgess.github.io/ai-security-wiki/wiki/risks/7.3.md) — ML Supply chain vulnerabilities
- [DASF-R-7.4](https://mathborgess.github.io/ai-security-wiki/wiki/risks/7.4.md) — Source code control attack
- [DASF-R-8.1](https://mathborgess.github.io/ai-security-wiki/wiki/risks/8.1.md) — Model attribution
- [DASF-R-8.2](https://mathborgess.github.io/ai-security-wiki/wiki/risks/8.2.md) — Model theft
- [DASF-R-8.4](https://mathborgess.github.io/ai-security-wiki/wiki/risks/8.4.md) — Model inversion
- [DASF-R-9.1](https://mathborgess.github.io/ai-security-wiki/wiki/risks/9.1.md) — Prompt inject
- [DASF-R-9.2](https://mathborgess.github.io/ai-security-wiki/wiki/risks/9.2.md) — Model inversion
- [DASF-R-9.3](https://mathborgess.github.io/ai-security-wiki/wiki/risks/9.3.md) — Model breakout
- [DASF-R-9.4](https://mathborgess.github.io/ai-security-wiki/wiki/risks/9.4.md) — Looped input
- [DASF-R-9.5](https://mathborgess.github.io/ai-security-wiki/wiki/risks/9.5.md) — Infer training data membership
- [DASF-R-9.7](https://mathborgess.github.io/ai-security-wiki/wiki/risks/9.7.md) — Denial of Service (DoS)
- [DASF-R-9.8](https://mathborgess.github.io/ai-security-wiki/wiki/risks/9.8.md) — LLM hallucinations
- [DASF-R-9.10](https://mathborgess.github.io/ai-security-wiki/wiki/risks/9.10.md) — Accidental exposure of unauthorized data to models
- [DASF-R-9.11](https://mathborgess.github.io/ai-security-wiki/wiki/risks/9.11.md) — Model Inference API Access
- [DASF-R-9.12](https://mathborgess.github.io/ai-security-wiki/wiki/risks/9.12.md) — LLM Jailbreak
- [DASF-R-9.13](https://mathborgess.github.io/ai-security-wiki/wiki/risks/9.13.md) — Excessive agency
- [DASF-R-10.1](https://mathborgess.github.io/ai-security-wiki/wiki/risks/10.1.md) — Lack of audit and monitoring inference quality
- [DASF-R-10.2](https://mathborgess.github.io/ai-security-wiki/wiki/risks/10.2.md) — Output manipulation
- [DASF-R-10.5](https://mathborgess.github.io/ai-security-wiki/wiki/risks/10.5.md) — Black-box attacks
- [DASF-R-10.6](https://mathborgess.github.io/ai-security-wiki/wiki/risks/10.6.md) — Sensitive data output from a model
- [DASF-R-11.1](https://mathborgess.github.io/ai-security-wiki/wiki/risks/11.1.md) — Lack of MLOps — repeatable enforced standards
- [DASF-R-12.1](https://mathborgess.github.io/ai-security-wiki/wiki/risks/12.1.md) — Lack of vulnerability management
- [DASF-R-12.2](https://mathborgess.github.io/ai-security-wiki/wiki/risks/12.2.md) — Lack of penetration testing, red teaming and bug bounty
- [DASF-R-12.3](https://mathborgess.github.io/ai-security-wiki/wiki/risks/12.3.md) — Lack of incident response
- [DASF-R-12.4](https://mathborgess.github.io/ai-security-wiki/wiki/risks/12.4.md) — Unauthorized privileged access
- [DASF-R-12.5](https://mathborgess.github.io/ai-security-wiki/wiki/risks/12.5.md) — Poor security in the software development lifecycle
- [DASF-R-12.6](https://mathborgess.github.io/ai-security-wiki/wiki/risks/12.6.md) — Lack of compliance
- [DASF-R-12.7](https://mathborgess.github.io/ai-security-wiki/wiki/risks/12.7.md) — Initial Access