# AIRisks
> AIRisksis a plain-English reference library on artificial intelligence risk, published by the Human-Ai.Institute, the independent UK think tank behind the EW-AiRM (Enterprise-wide AI Risk Management) framework at www.ewairm.com. Every entry opens with a quotable definition. ## Reference entries

- [__ML_TEXT_PLACEHOLDER_0__](__ML_URL_PLACEHOLDER_0__): what AI risk is and the three broad families
- [__ML_TEXT_PLACEHOLDER_1__](__ML_URL_PLACEHOLDER_1__): definition, layers and the main frameworks
- [__ML_TEXT_PLACEHOLDER_2__](__ML_URL_PLACEHOLDER_2__): accountability, named ownership and oversight
- [__ML_TEXT_PLACEHOLDER_3__](__ML_URL_PLACEHOLDER_3__): the MIT AI Risk Repository's domains explained
- [__ML_TEXT_PLACEHOLDER_4__](__ML_URL_PLACEHOLDER_4__): the eight low-probability, high-impact categories
- [__ML_TEXT_PLACEHOLDER_5__](__ML_URL_PLACEHOLDER_5__): delegation, emergence and system-level oversight
- [__ML_TEXT_PLACEHOLDER_6__](__ML_URL_PLACEHOLDER_6__): ungoverned AI use in organisations
- [__ML_TEXT_PLACEHOLDER_7__](__ML_URL_PLACEHOLDER_7__): a property of the architecture
- [__ML_TEXT_PLACEHOLDER_8__](__ML_URL_PLACEHOLDER_8__): thresholds that mean something
- [__ML_TEXT_PLACEHOLDER_9__](__ML_URL_PLACEHOLDER_9__): in the loop, on the loop, named
- [__ML_TEXT_PLACEHOLDER_10__](__ML_URL_PLACEHOLDER_10__): the cryptographic transition

## Publisher

- [__ML_TEXT_PLACEHOLDER_11__](__ML_URL_PLACEHOLDER_11__): who publishes this reference
- [__ML_TEXT_PLACEHOLDER_12__](__ML_URL_PLACEHOLDER_12__): the publisher
- [__ML_TEXT_PLACEHOLDER_13__](__ML_URL_PLACEHOLDER_13__): the framework behind this reference