AI Models Trained on Incomplete Data Can’t Protect Against Threats

In cybersecurity, AI is being called the future of threat finder. However, AI has its hands tied, they are only as good as their data pipeline. But this principle is not stopping at academic machine learning, as it is also applicable for cybersecurity.

AI-powered threat hunting will only be successful if the data infrastructure is strong too.

Threat hunting powered by AI, automation, or human investigation will only ever be as effective as the data infrastructure it stands on. Sometimes, security teams build AI over leaked data or without proper data care. This can create issues later. It can affect both AI and humans. Even sophisticated algorithms can’t handle inconsistent or incomplete data. AI that is trained on poor data will also lead to poor results. 

The importance of unified data 

A correlated data controls the operation. It reduces noise and helps in noticing patterns that manual systems can’t.

Correlating and pre-transforming the data makes it easy for LLMs and other AI tools. It also allows connected components to surface naturally. 

A same person may show up under entirely distinct names as an IAM principal in AWS, a committer in GitHub, and a document owner in Google Workspace. You only have a small portion of the truth when you look at any one of those signs. 

You have behavioral clarity when you consider them collectively. While downloading dozens of items from Google Workspace may look strange on

[…]
Content was cut in order to protect the source.Please visit the source for the rest of the article.

This article has been indexed from CySecurity News – Latest Information Security and Hacking Incidents

Read the original article: