Cambridge, Massachusetts / Access to Newswire / 29 September 2025 / Cybera, a global leader in SCAM Prevention and Mule Account Intelligence, has announced today that it has been accepted by rigorous peer review to present its latest research paper: “Send to which account? Evaluation of an LLM-based scambaut system”, at the APWG Ecrime 2025 Conference In San Diego in November.
Scammers are increasingly using generative AI (Genai) technologies to produce very convincing phishing capacity on a scale, to stimulate financial fraud and eradicate trust. Cybera’s research demonstrates a new, proactive approach: fascinating scammers directly through conversation -HonePots driven by large language models (LLMS) to extract usable threat intelligence.
During a five -month deployment, the Cybera system initiated more than 2,600 conversations with real scammers, who generated more than 18,700 messages. The system achieved an information -public prosecutor percentage (IDR) of 32%, successfully sensitive financial details such as Mule Bank accounts and cryptocurrency portfolios. This intelligence is crucial for banks, payment processors and law enforcement to freeze funds and protect consumers before losses take place.
“At Cybera we believe that AI should not only protect people against scams, but must actively disrupt the criminal infrastructure that makes fraud possible,” said Cybera CEO Patrick Peterson. “Our LLM-based scambaiting system represents a new limit in proactive fraud fee-one where we can collect intelligence on a scale and share it with financial institutions to stop scam before victims lose money.”
The research team included Hossein Siadati (Cybera), Haadi Jafarian (University of Colorado Denver) and Sima Jafaikhah (UNCW). Their work not only demonstrates the potential of AI-driven involvement in improving fraud prevention, but also identifies operational challenges, including the need to improve the initial scammer engine figures, offering guidelines for the next generation of automated defenses.
“The collaborations of industrial laboratories and academic research centers are a characteristic of research presented on APWG Ecrime Symposia,” said APWG -Secretary -General and founder of ECRIME Symposium Peter Cassidy. “Again, in this groundbreaking work by Siadati, Jafarian and Jafaikhah, we see the powerful research insights arising from collaborations in the sector.
The APWG Ecrime conferencethis year held from 4-7 November on the Loews Coronado Bay Resortis a prominent worldwide event that brings together researchers, financial institutions, law enforcement and security practitioners to tackle the global cyber crime problem. The Peer-Reviewed, Publishing Conference (Proceedings Published by IEEE Since 2008) is celebrating its 20th birthday edition in November.
About Cybera
Cybera is a global Mule Intelligence and Fraud Prevention Company that helps banks, payment providers and supervisors to protect customers against scams. By combining ethical scambanes, AI-driven intelligence collection and real-time data exchange, enables Cybera institutions to block fraudulent transactions, freeze mules and reduce the financial and emotional impact of fraud worldwide.
About APWG
The APWG (Anti-Phishing Working Group) is an international coalition of responds of counter-cyber crime, forensic researchers, law enforcement agencies, technology companies, financial service providers, retailers and brand holders, university researchers, NGOs and multilateral treaty organizations. The directors, managers and research smartows advise national and subnational governments and the United Nations (Office on Drugs and Crime) as recognized experts (under the Doha statement of 2010 and Salvador Declaration of 2015), as well as multilateral treaty institutions and industrial organizations. APWG forges worldwide recognized conventions in data exchange and manages applied research, industrial policy and public awareness reports at work all over the world.
Mediacontact
Claudio Staub
CRO, co-founder
[email protected]
+1 650 822 6065
APWG: [email protected]
SOURCE: Anti-Phishing Working Group
