# Trust is measured, not claimed.

> The Trust and Safety Institute (TSI) is an independent nonprofit setting trust and safety standards for AI agents: TSI defines the standard and issues the score, AISquare produces the evidence.

- **URL:** https://aisquare.studio/tsi
- **Audience:** Enterprises that need their AI agents independently assessed, and anyone checking how AISquare relates to TSI
- **Topics:** Trust and Safety Institute, TSI, TSI Score, TSI Trust Tag, TSI Trust Stack, trust and safety standards for AI agents, independent assessment, White House Accord on Super Intelligence, agent trust, signed evidence, RML

## Trust and Safety Institute

The Trust and Safety Institute (TSI) is an independent nonprofit setting trust and safety standards for AI agents. TSI defines the standard and issues the score. AISquare is a separate platform that produces the evidence.

## TSI Trust Tag

A3 Assured. Independently assessed against global standards and the TSI Trust Stack.

- A1 Foundational

- A2 Developing

- A3 Assured

- A4 Advanced

- A5 Trusted at Scale

Evidence from AISquare. Signed runtime records · RML.

Sample Trust Tag. Ratings are issued by TSI under its own methodology.

## The White House Accord asks for independent assessment.

Signed on September 29, 2026 by the President and the leaders of Google, Anthropic, Meta, OpenAI, xAI and Nvidia, the Accord asks every company training and deploying frontier models for four layers of controls and audits.

- Internal controls and monitoring

- An internal team that makes sure the controls work

- An independent external auditor or evaluator (The role of an independent institute)

- An independent board committee that oversees it all (The role of an independent institute)

The Accord is voluntary and written for frontier AI labs. Enterprises putting agents to work face the same question from their boards, customers and regulators: who checked?

White House Accord on Super Intelligence, September 29, 2026. Source: The American Presidency Project.

The signatories of the Accord are not affiliated with TSI or AISquare and do not endorse them.

## TSI sets the bar. AISquare produces the evidence.

- Capture data (AISquare can supply this)

- Store evidence (AISquare can supply this)

- Map to standards (AISquare can supply this)

- Define tests (AISquare can supply this)

- Continuously evaluate (AISquare can supply this)

- Create verifiable evidence (AISquare can supply this)

- Issue the Trust Tag (Issued by TSI)

TSI: an independent institute and standards authority. Defines the Trust Stack and issues the TSI Score and Trust Tag.

AISquare: a separate enterprise platform. Records how agents decide, enforces your rules and produces signed evidence.

We're working with the Trust and Safety Institute to publish RML as an open standard.

## Help shape how agent trust is measured.

- Get your TSI Score

- Join a working group

- Join the AI Debrief Room

## Show your agents can be trusted.

Run them on AISquare. Get them assessed by TSI.

## Links

- [Visit the Trust and Safety Institute](https://trustandsafety.institute)
- [Get your TSI Score](https://trustandsafety.institute/tsi-score)
- [Join a working group](https://trustandsafety.institute/#groups)
- [Join the AI Debrief Room](https://trustandsafety.institute/debrief-room)
- [Download the Accord (PDF)](https://aisquare.studio/tsi/white-house-accord-super-intelligence.pdf)
- [See RML](https://aisquare.studio/product/reasoning-graph)
- [Book a demo](https://calendly.com/d/dws4-msf-8jc/aisquare-gtm-team)

---

**Machine-readable AISquare**

- Site index for LLMs: https://aisquare.studio/llms.txt
- Agent instructions (when to use AISquare): https://aisquare.studio/agents.md
- OpenAPI specification: https://aisquare.studio/openapi.json
- MCP endpoint (Streamable HTTP, POST): https://aisquare.studio/mcp
- Sitemap: https://aisquare.studio/sitemap.xml
- Blog feed (RSS, full posts): https://aisquare.studio/blogs/rss.xml
- Any page on this site also serves markdown: append `.md` to its path (the homepage is `/index.md`) or send `Accept: text/markdown`.
