Every day, millions of consumers feel like their phones are ringing with another unwanted call, a problem that continues to grow despite years of regulation and filtering tools. According to the U.S. Public Interest Research Group, Americans receive 29.6 billion robocalls in 2025, showing how persistent and evolving this problem is. What seems like a random nuisance is increasingly supported by organized infrastructure, where large-scale networks, not individual actors, drive large-scale fraud.
SIM Farm Rise
At the center of modern spam operations Sim Farmslarge groups of real SIM cards connected to devices that can make thousands of calls simultaneously. Since these calls originate from legitimate numbers and mimic normal user behavior, they are difficult to detect using traditional filters.
Like the FCC NotesMany spam calls exploit vulnerabilities in communications authentication systems, allowing bad actors to operate within the same networks used for legitimate communications.
This infrastructure has transformed spam from a nuisance into an industrial system. Instead of relying on a single number or text, operators can rotate through thousands of SIM cards, distributing activity and adapting tactics in real time.
Advances in artificial intelligence are amplifying this shift. like Covered by Mashablescammers are increasingly using AI-generated voices to make calls more convincing, blurring the line between automated systems and human interaction.
The result is a structural challenge for telecommunications service providers. Traditional spam detection systems rely heavily on fixed rules, such as identifying known bad numbers or flagging unusual call volumes. SIM farms break this model by distributing activity across many numbers that each look normal in isolation. From a network perspective, traffic often appears legitimate, making it difficult to distinguish between real users and coordinated fraud.
Advertisement: Scroll to continue
From filtering to modeling
We would love to be yours Favorite source of news.
Please add us to your favorite sources list so our news, statements and interviews appear in your feed. Thanks!
New research from Virginia TechIt signals a change in how the industry is responding. Instead of relying solely on interactive filtering, researchers are using AI to model and detect curated SIM farm activity by analyzing behavioral patterns across large volumes of calls.
The main innovation is the use of digital twins for communications networks. This simulated environment mirrors real-world network behavior, allowing researchers to recreate how SIM farms operate on a large scale. Within this controlled setting, AI systems can be trained to identify patterns that indicate coordinated fraud, such as simultaneous dialing behavior, unusual routing patterns, or rapid switching between SIM cards.
This approach addresses one of the fundamental limitations in telecom fraud detection: access to data. As noted in Virginia Tech Research,Telecom operators closely protect customer data and network ,information, making it difficult for outside researchers to test ,detection systems in real-world conditions. The digital twin provides an alternative solution by enabling realistic simulation without revealing sensitive data.
AI is also being deployed operationally by telecommunications service providers. like Covered by PYMNTSAT&T uses autonomous AI agents to detect fraud, manage network anomalies and reduce response times. These systems analyze massive amounts of network data in real-time, allowing for faster identification of suspicious activity and more adaptive defenses.
Blocking limits
Despite these advances, stopping unwanted calls remains a major challenge. Consumer-facing solutions, such as call blocking apps and device-level filters, provide some relief but are limited in scope. like Covered by CNETEven the most effective tools often rely on user reports and known spam databases, which can lag behind rapidly evolving tactics.
The biggest problem is that communications networks were not designed with competitive AI in mind. Authentication frameworks, numbering systems, and routing protocols assume a level of trust that modern fraud exploits. As a result, defenses that focus on blocking individual calls or numbers are reactive in nature.
AI changes the equation by allowing a more preventative approach. Instead of chasing individual spam calls, systems can analyze network-level behavior, identify coordinated activity, and intervene early in the attack lifecycle. The use of simulated environments enhances this capability by allowing defenses to be tested and improved against evolving tactics.
For all of our PYMNTS AI coverage, subscribe to our daily newsletter Amnesty International newsletter.
