AI Agents Go Undercover to Outsmart Scammers
Most companies harness artificial intelligence to accelerate product delivery or streamline workflows. However, Apate.ai, an innovative Australian startup founded in 2024, uses AI agents in a novel and unexpected way: by sending them undercover to get scammed. This approach flips the traditional narrative of AI as a mere tool for efficiency, transforming it into a frontline defense against cybercriminals.
Originating from research conducted at Macquarie University in Sydney, Apate.ai deploys thousands of AI agents that impersonate unsuspecting victims. These agents engage with scammers by answering calls, responding to dubious messages and emails, clicking phishing links, and even infiltrating social media groups designed to bait fraudsters. Their personas vary widely, reflecting different ages, genders, accents, and languages, enabling them to maintain prolonged conversations that extract valuable intelligence.
Turning the Tables on Scammers
Apate’s fleet of agents has already wasted an estimated 17 years of scammers’ time through 2.5 million conversations, according to company data. More critically, these “undercover agents” coax scammers into revealing confidential information typically uncovered only after a victim falls prey: bank account details, cryptocurrency wallet addresses, phone numbers, malicious URLs, and even the brands and tactics fraudsters impersonate.
Serving a diverse client base—including banks, telecommunications companies, government agencies, and utilities—Apate delivers this intelligence to help organizations bolster their anti-fraud measures. The company was recently named a finalist in the Startup Battlefield 200 at TechCrunch Disrupt 2026, where it is pitching to venture capital judges in San Francisco this October.
The Growing Threat of AI-Enabled Scams
The rise of AI has not only empowered legitimate businesses but also scammers and hackers. The FBI’s 2025 Internet Crime Report reveals over 22,000 complaints of AI-related crimes resulting in nearly $893 million in losses. Investment scams, many leveraging AI to simulate conversations or deepfake trusted figures’ voices and appearances, accounted for more than $632 million of these losses.
Historically, some individuals have entertained themselves by wasting scammers’ time—famously coaxing them into reading entire books aloud or even getting absurd tattoos. With AI’s increasing sophistication, companies like Apate are professionalizing this scam-baiting, turning it into a strategic intelligence-gathering operation.
From Personal Experience to Tech Innovation
Apate’s founder and CEO, Dali Kaafar, attributes the company’s inception to a personal experience. During a family picnic, Kaafar received a scam call and deliberately prolonged the conversation for 44 minutes to protect his children from exposure. This sparked the idea that such interactions could be automated, simultaneously wasting scammers’ time and gathering useful data.
Building on this concept, Kaafar and his colleagues at Macquarie University applied machine learning and natural language processing techniques to analyze scam calls and develop enhanced detection and prevention tools. This academic foundation underpins Apate’s cutting-edge technology today.
How Apate’s AI Agents Work
Brad Joffe, Apate’s co-founder and chief commercial officer, explains that while many organizations react after a scam occurs—blocking numbers or monitoring transactions—Apate proactively engages scammers directly. “Our AI agents talk to tens of thousands of scammers every day across calls, messages, email, and the web. What we offer our customers is what comes out of those conversations, and what they can do with it,” Joffe told TechCrunch.
The company utilizes both state-of-the-art frontier AI models and proprietary fine-tuned models trained on scammer interactions and academic research. This combination allows Apate agents to mimic human-like conversation effectively, keeping scammers engaged long enough to extract critical insights.
Customers gain access not only to Apate’s agent network but also to an intelligence portal providing analytics on scam trends targeting their users, employees, brands, and products. Businesses can assign specific missions to agents, such as obtaining bank account or crypto wallet numbers. Additionally, text-based agents scour forums, investment groups, and social platforms, adopting fabricated identities to infiltrate scammer networks and gather actionable data.
Proven Impact and Growing Momentum
Apate’s client list includes major players like Commonwealth Bank of Australia and TPG Telecom, alongside various utilities and government bodies. TPG has rerouted nearly 590,000 scam calls to Apate’s AI agents, while Commonwealth Bank has reported a 76% reduction in publicly disclosed scam losses. Another unnamed Australian bank leveraged Apate’s outbound agents to disrupt 22 scam campaigns, demonstrating the platform’s efficacy.
Investor confidence is strong. In August 2026, Apate secured $8.15 million in seed funding from prominent firms including Lobby Capital, OIF Ventures, Investible, Concept Ventures, and Baobab Ventures. The company, now 20 employees strong with nearly $10 million raised in total, plans to expand internationally, establishing a presence in London and increasing its footprint in the U.S.
A Name Rooted in Deception
The company’s name, Apate, derives from the Greek goddess of deception—a fitting emblem for AI agents designed to trick scammers. Joffe notes with humor that the startup celebrates when scammers swear at their bots, interpreting profanity as a sign of frustration and engagement. “We are probably one of the only businesses in the world that celebrates when people swear at our product,” he said.
By leveraging AI not just to streamline business processes but to actively combat fraudsters, Apate.ai exemplifies how technology can be repurposed for social good, protecting individuals and organizations from increasingly sophisticated scams.
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.
Source: Here
