Apate.ai Deploys AI Agents to Infiltrate Scammers and Feed Fraud Intelligence to Banks
The Australian startup uses thousands of AI bots to pose as victims, extracting scammer details for banks, telcos and governments.
Apate.ai, an Australian company founded in 2024 on research from Macquarie University, has built a fleet of artificial-intelligence agents that go undercover with scammers. The bots answer phone calls, reply to suspicious messages, follow phishing links and join social-media groups, all while pretending to be ordinary people. The startup is a finalist in this year’s Startup Battlefield 200 at TechCrunch Disrupt in San Francisco.
The agents are programmed to adopt a wide range of ages, genders, accents and languages, allowing them to stay engaged with fraudsters long enough to collect actionable data. According to the company, the bots have already logged 2.5 million conversations, wasting an estimated 17 years of scammers’ time. More importantly, the interactions reveal details that victims typically learn only after a loss, such as bank account numbers, cryptocurrency wallet addresses, phone numbers, malicious URLs and the brands being impersonated.
Apate’s customers include the Commonwealth Bank of Australia, TPG Telecom and a number of utilities, telecommunications firms and government agencies. TPG has redirected nearly 590 000 calls to the AI agents, while Commonwealth reported a 76 percent drop in publicly disclosed scam losses after using the service. An unnamed Australian bank said the outbound agents helped disrupt 22 separate scam campaigns.
The technology combines leading-edge large language models with proprietary fine-tuned versions. Training data comes from the conversations the agents have with fraudsters as well as earlier research conducted by founder Dali Kaafar at Macquarie. Clients receive access to an insights portal that aggregates the extracted intelligence and offers analytics on threats targeting their users, brands or products. The platform also lets customers assign specific missions, such as obtaining a victim’s bank details or locating a crypto wallet, and deploy text-based agents that can infiltrate forums and investment groups.
Kaafar’s idea originated from a personal experience at a family picnic when he kept a scam call alive for 44 minutes to entertain his children. The wasted time sparked the question of whether a machine could do the same at scale and return useful information. That curiosity led to academic work on machine-learning techniques for decoding scam calls, which later formed the foundation of Apate’s product.
In August, Apate closed an $8.15 million seed round led by Lobby Capital, OIF Ventures, Investible, Concept Ventures and Baobab Ventures. The 20-person company has raised just under $10 million in total and plans to allocate the new capital to product development and international expansion, including a new office in London and a larger footprint in the United States.
The startup’s model arrives as AI-enabled fraud is on the rise. The FBI recorded more than 22 000 complaints of AI-related crimes in 2025, resulting in roughly $893 million in losses, with investment scams accounting for $632 million of that total. By turning AI against the fraudsters themselves, Apate offers a proactive alternative to the traditional reactive defenses used by banks and telcos.
Looking ahead, the company aims to scale its agent network and refine its analytics, while navigating regulatory scrutiny around automated interactions and data privacy. Whether AI can stay ahead of increasingly sophisticated scammers remains an open question, but Apate’s early results suggest that confronting fraud at its source could become a new pillar of anti-fraud strategy.