Points of Focus

• TRM Labs launched Co-Case Agent to track fast-moving crypto crime.
• Illicit crypto hit $158B in 2025; AI scams jumped 500% YoY.
• The agent speeds up probes while keeping them transparent and human-led.

Criminal networks can now move and obscure funds across chains faster than investigators can track even a single hop. TRM Labs, the blockchain intelligence firm operating across 50+ countries, launched Co-Case Agent on March 25, 2026, to address this exact challenge. 

It’s an AI-powered investigative assistant embedded within TRM Forensics, its flagship investigations tool. The agent converts plain-language queries into multi-step on-chain investigative actions: tracing fund flows within seconds, running on-demand graph audits, providing next-step recommendations, and logging every interaction to an immutable audit trail.

Meet 𝐂𝐨-𝐂𝐚𝐬𝐞 𝐀𝐠𝐞𝐧𝐭™, an AI assistant for TRM Forensics customers that helps teams:

▪️ Trace fund flows in seconds
▪️ Ensure investigative defensibility
▪️ Sharpen investigative tradecraft
▪️ Stay in control of every decision

Available now at no additional cost.… pic.twitter.com/3XPgoxrLqt

— TRM Labs (@trmlabs) March 25, 2026

Co-Case Agent can be accessed immediately at no additional cost by all existing TRM Forensics customers, spanning law enforcement agencies, financial institutions, regulatory bodies, and national security agencies. The launch is a direct response to a widening gap between investigator capacity and the growing scale of crypto-related crime. 

TRM’s 2026 Crypto Crime Report put total illicit on-chain crypto volume at $158 billion in 2025, with scam proceeds accounting for approximately $30 billion and hack-related theft adding another $2.87 billion. AI-enabled scams alone surged 500% year-over-year, powered by criminal networks using deepfakes, automated social engineering, and cross-chain fund movement that leaves investigators with almost no time to intervene.

At the same time, investigators are reportedly being asked to simultaneously manage activity across dozens of blockchains, jurisdictions, and criminal typologies, with caseloads growing faster than the workforce can scale.

🤖TECH: @TRMLABS OFFERS NEW AI AGENT PROGRAM TO HELP CATCH CRYPTO CRIMINALS

New investigative AI agents are embedded in a TRM Forensics service that's extended to law enforcement agencies, crypto businesses and financial firms TRM said in a press release.

A user can request… pic.twitter.com/0byP1sUNzn

— BSCN (@BSCNews) March 25, 2026

A parallel agent running on every active case

Co-Case Agent is architected to run concurrently across an investigator’s entire caseload. At each stage of an investigation, it traces funds, proposes the next logical step, and records every action to an immutable audit log. This creates a complete chain-of-custody record from the first prompt to the final export. That record is designed to hold up whether the output is a SAR filing, a court exhibit, or a response to a regulatory examination.

TRM Labs says the agent applies investigator-level expertise faster and at a scale far beyond what manual workflows can achieve, effectively acting as a parallel assistant on every active case.

Glass-box auditability in an increasingly opaque threat environment

While most AI tools prioritize speed over transparency, Co-Case Agent takes a different approach. TRM Labs describes it as a “glass-box” philosophy, where every recommendation is fully traceable to underlying transaction graphs and attribution data. The agent can suggest, explain, and document findings, but never alters data or draws conclusions without a clear, auditable basis.

Informed decisions on blockchain intelligence require transparency and reliability. TRM Labs provides our customers with "glass box" attribution, ensuring full transparency of data sources. Learn more ➡️ https://t.co/LsTkKAGrCR #blockchain #compliance #lawenforcement pic.twitter.com/XOYVzh59Cr

— TRM Labs (@trmlabs) May 28, 2024

That design carries direct legal weight. Courts and regulators require full methodological transparency, and an AI output that cannot be traced, explained, and defended in cross-examination creates evidentiary exposure rather than investigative value. 

A March 19, 2026, TRM analysis on AI in crypto investigations emphasized that the reasoning chain, not just the algorithmic output, is what forms the evidentiary record in legal proceedings.

That study further stresses that AI accelerates data triage, clustering, and pattern detection but stops short of legal interpretation. Investigators still evaluate context, establish intent, and determine accountability, steps that algorithms alone cannot handle. Hence, humans retain final control over analysis, documentation, and decision-making.

Additional Co-Case Agent capabilities are expected to roll out gradually.