Human traders or analysts review potentially suspicious activity to:

1. Contextualize alerts: Assess the market context and trading activity to determine if the alert is legitimate.

2. Investigate further: Analyze trade data, market conditions, and other factors to determine if manipulation has occurred.

3. Exercise judgment: Use expertise and experience to make informed decisions about whether activity is indeed manipulative.

4. Escalate incidents: Refer cases to regulatory authorities or internal teams for further investigation and action.

5. Refine rules and models: Use human oversight findings to improve and refine rule-based systems and machine learning models.

6. Identify new threats: Detect emerging manipulation tactics and strategies that may not be caught by automated systems.

7. Provide expertise: Offer insights and recommendations to improve market integrity and prevent manipulation.

8. Enhance transparency: Provide clear explanations for trading halts or other actions taken to prevent manipulation.

Human oversight adds a critical layer of nuance and expertise to AMMs' surveillance capabilities, helping to:

1. Reduce false positives

2. Improve detection accuracy

3. Enhance market trust

4. Support regulatory compliance

5. Foster a fair and efficient market environment

By combining human oversight with automated surveillance systems, AMMs can create a robust and effective framework for preventing market manipulation.