diff --git a/content/ai_exchange/content/docs/3_development_time_threats.md b/content/ai_exchange/content/docs/3_development_time_threats.md
index 7909019..a595295 100644
--- a/content/ai_exchange/content/docs/3_development_time_threats.md
+++ b/content/ai_exchange/content/docs/3_development_time_threats.md
@@ -332,6 +332,9 @@ Key Points for Consideration:
 - Continuous Monitoring: Regularly update and audit data quality controls to adapt to evolving threats and maintain the robustness of AI systems.
 - Collaboration and Standards: Adhere to international standards like ISO/IEC 5259 and 42001 while recognizing their limitations. Advocate for the development of more comprehensive standards that address the unique challenges of AI data quality.
 
+References
+- ['Detection of Adversarial Training Examples in Poisoning Attacks through Anomaly Detection'](https://arxiv.org/abs/1802.03041)
+
 Useful standards include:
 
 - ISO/IEC 5259 series on Data quality for analytics and ML. Gap: covers this control minimally. in light of the particularity - the standard does not mention approaches to detect malicious changes (including detecting statistical deviations). Nevertheless, standard data quality control helps to detect malicious changes that violate data quality rules.