AI is now screening prison communications to forecast crimes

Could artificial intelligence one day predict criminal activity before it occurs? A company says its AI system might be able to do just that.

Telecom firm Securus Technologies claims it has developed an AI model trained on years of text, phone, and video communications from prison inmates, designed to detect signs of imminent criminal activity. The tool has been in early use over the past year.

In an interview with MIT Technology Review, Securus president Kevin Elder said the AI can monitor communications in real time and identify “when crimes are being thought about or contemplated.”

Securus says the system can be tailored for specific jurisdictions, citing a version trained exclusively on calls from inmates in Texas. The AI is meant to flag suspicious activity for human agents, who then review the content more closely.

The company reports that its tools have contributed to stopping human trafficking, smuggling, and gang-related activity, though it has not provided detailed examples or revealed the current locations of its deployments.

Privacy advocates have raised concerns, noting that while inmates’ calls are routinely recorded, they have not consented to their data being used to train AI systems.

Recent policy changes have also shifted some AI development costs onto prisoners. In June 2025, FCC Commissioner Brendan Carr announced a delay on new caps for prison phone rates, paving the way for expanded use of AI tools in correctional settings.

“With today’s actions, we’re helping to ensure that communications are more readily available and that important safety and security protocols are maintained. This includes steps that can lead to broader adoption of beneficial public safety tools that include advanced AI and machine learning,” Carr wrote.

Original Source