Handshake AI’s work depends on trusted, high-quality human expertise. To protect the integrity of our projects and ensure fairness for all fellows, all work submitted must be your own original effort and created without the use of external AI tools.

These standards help us maintain trust with our partner labs, ensure consistent quality, and make sure fellows are evaluated and paid fairly for their own work. The policy below explains what’s required and how to stay in good standing.

What does it mean to create original work product for Handshake AI?

Handshake AI aims to help shape the AI industry through high-quality, trusted human intelligence.  Your contributions must be original and created solely by you.  

For clarity, the following examples are not original contributions:

• Prompts or questions taken from resources like Khan Academy or Chegg;
• Prompts from textbooks;
• Prompts from your annotation work on other platforms. 

Can I use LLMs to help me create work product for Handshake AI?

Generally, no.  You are strictly prohibited from using Large Language Models or other artificial intelligence‑assisted tools (collectively, “LLMs”) to generate work product for Handshake AI, unless otherwise expressly authorized in writing by Handshake. 

LLM usage includes:

• developing or refining prompt ideas, 
• generating or editing content, 
• solving tasks, or
• using LLM-based grammar and writing tools for stylistic improvements, grammar correction, or formatting. 

Only use the Handshake AI platform or tools expressly authorized in writing by Handshake to generate work product.

What records should I keep of my effort?

You must retain all preparatory materials used to generate finished work product for at least 90 days after you deliver the work product to Handshake. 

Preparatory materials include:

• draft prompts and annotations,
• notes, 
• ideas and 
• other written documents.

Handshake may require you to provide copies of such preparatory materials for its review.

What are the consequences of a breach of the LLM usage prohibition?

If Handshake believes that you have used an LLM to generate work product, all payments to you for such Work Product will be suspended pending Handshake’s further review. Handshake may provide you with an opportunity to provide evidence that you did not utilize an LLM to generate the applicable work product. 

If Handshake concludes that you used an LLM to generate the applicable work product:

• it will reject the work product and you will not be paid for it (including for any time spent generating that work product), and
• Handshake may offboard you from the applicable Project or terminate you from the Program.

Please refer to section 7 of the Contractor Agreement for more information about the prohibition on LLM usage.