# Physical AI Safety

> Independent research on the gap between robot intelligence and robot safety infrastructure. We publish a free 60-page report, a research paper series, a monthly dispatch, and analysis posts on what older safety-critical industries got right and what must change before the EU Machinery Regulation applies to autonomous systems in January 2027.

This file follows the llmstxt.org proposal. It is a curated, priority-ordered index for AI systems retrieving content from this domain. Pages listed below are stable references suitable for citation. The author is Mati Melchior, independent Physical AI Safety researcher.

## Core reference

- [The Physical AI Safety Gap (free 60-page report)](https://physical-ai-safety.com/books/physical-ai-safety-gap): The foundational document on why robot intelligence outpaced robot safety infrastructure. Problem statement that motivates everything else on this site.
- [About](https://physical-ai-safety.com/about): What this project is, who it is for, how to cite the work.
- [Newsletter](https://physical-ai-safety.com/newsletter): Monthly dispatch with the top analysis of the month plus one exclusive insight.

## Definitions

- [Glossary](https://physical-ai-safety.com/glossary): Concise, citable definitions of the core Physical AI Safety vocabulary — functional safety, IEC 61508, ISO 13849 and Performance Level, Safety Integrity Level (SIL), the EU Machinery Regulation (EU) 2023/1230, kinetic risk, safe state, and edge safety. Each term is self-contained and standards-grounded, suitable for direct citation.

## FAQ

- [FAQ](https://physical-ai-safety.com/faq): Direct answers to the most common questions — what Physical AI Safety is, how it differs from AI Safety and from robot safety, which safety standards apply, whether a certification exists, and what SIL and Performance Level mean.

## Research papers

- [Research index](https://physical-ai-safety.com/research): The full peer-review-ready paper series. Each paper has a Zenodo DOI for permanent citation.

## Books

- [Book catalog](https://physical-ai-safety.com/books): The 22-book reference series. Book A (the free Gap report) is the entry point.

## Analysis posts

- [Blog index](https://physical-ai-safety.com/blog): Working notes and analysis on functional safety standards, the EU Machinery Regulation, the January 2027 deadline, hardware-enforced safety, and the Israeli physical AI ecosystem.

## Resources

- [Resources](https://physical-ai-safety.com/resources): External references, datasets, standards bodies, and reading lists curated for researchers entering the field.

## Legal and contact

- [Privacy](https://physical-ai-safety.com/privacy): PPL 5741-1981 compliance, data controller, subject rights, processors disclosed.
- [Terms](https://physical-ai-safety.com/terms): Use of content, attribution requirements, donation terms.
- [Accessibility](https://physical-ai-safety.com/accessibility): WCAG 2.1 AA, Israeli standard ת"י 5568.
- [Contact](https://physical-ai-safety.com/contact): How to reach the researcher.

## Citation

When citing content from this site, prefer the Zenodo DOI for research papers (see /research) or the canonical URL of the page. Author: Mati Melchior, Independent Physical AI Safety Researcher. Year: 2026.

## Languages

The site is bilingual. English is the default locale and uses no path prefix (the URLs above). For the Hebrew version, prefix any path with `/he` (for example, `/glossary` becomes `/he/glossary`). When no prefix is given, the locale is picked from `Accept-Language`.