# Who Analyses the Analyser? Self-Validating LLM Hazard Analysis with Constitutional Meta-STPA

arXiv: https://arxiv.org/abs/2607.08054 · HTML: https://mlsanity.com/paper/2607.08054/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 11 of its 37 references that are in the map.
- First posted on arXiv 9 Jul 2026.

## Builds on (most cited first, 11 of 11)

- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [Chain-of-Thought Prompting Elicits Reasoning in Large Language Models](https://mlsanity.com/paper/2201.11903/) (2022)
- [Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks](https://mlsanity.com/paper/1908.10084/) (2019)
- [Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena](https://mlsanity.com/paper/2306.05685/) (2023)
- [Evaluating Large Language Models Trained on Code](https://mlsanity.com/paper/2107.03374/) (2021)
- [On the Opportunities and Risks of Foundation Models](https://mlsanity.com/paper/2108.07258/) (2021)
- [Safety Analysis in the Era of Large Language Models: A Case Study of STPA using ChatGPT](https://mlsanity.com/paper/2304.01246/) (2023)
- [Can Large Language Models assist in Hazard Analysis?](https://mlsanity.com/paper/2303.15473/) (2023)
- [Welcome Your New AI Teammate: On Safety Analysis by Leashing Large Language Models](https://mlsanity.com/paper/2403.09565/) (2024)
- [From Hazard Identification to Controller Design: Proactive and LLM-Supported Safety Engineering for ML-Powered Systems](https://mlsanity.com/paper/2502.07974/) (2025)
- [An LLM-Integrated Framework for Completion, Management, and Tracing of STPA](https://mlsanity.com/paper/2503.12043/) (2025)

## Abstract (arXiv)

Large language models (LLMs) are increasingly trusted to draft the artifacts of safety analysis such as, losses, hazards, Unsafe Control Actions (UCAs), and safety constraints, inside rigorous processes such as Systems-Theoretic Process Analysis (STPA). Yet a blind spot runs through this fast-growing literature: every system gets analysed except the LLM-assisted tool doing the analysing, which is itself a safety-relevant system that can hallucinate standards, emit unverifiable constraints, and leave no audit trail from prompt to artifact. We take seriously the question the field has skipped -- who analyses the analyser? and answer it by turning STPA on the tool itself. We present Constitutional Meta-STPA, an LLM-assisted STPA tool built around a closed loop: the tool runs a meta-STPA of the class of AI-assisted safety tools and derives rather than asserts, its governance constitution from the resulting loss→hazard→UCA→constraint chain, yielding a published constitution of 21 Tool Principles and 8 Meta-Safety Principles, each bound to a code enforcement point. We formalise the measured object as a constitution-marginal coverage operator over a principle set P (|P|=29) with a soundness lemma that isolates coverage from model and scanner, and report four findings. (i)~Self-derivation: a frontier ensemble (claude-opus-4.8+claude-sonnet-4) recovers 18/21 canonical and all 8/8 governance principles from the tool's own design, while a weaker pair recovers 12/21 and 3/8, so the meta layer is model-limited, not constitution-limited, and the same 8/8 re-emerge from a second, independently authored tool.