Generative AI has made fraud cheaper to run at scale. Deepfaked celebrities and automated AI conversations are now common tactics. Australian startup Apate is taking the same technology in the opposite direction, sending AI agents to pose as victims, answer scam calls, click phishing links, and collect intelligence while tying up fraud operations.
According to TechCrunch, Apate’s AI agents have participated in 2.5 million conversations so far, collectively wasting 17 years of scammers’ time. The report framed the issue plainly: where AI is deployed will shape the direction of the anti-scam fight.
Scammers are already using AI as an operating tool
TechCrunch said fraudsters use AI to lead conversations with targets and to deepfake the voices and appearances of celebrities, corporate executives, and other trusted figures, making false messages harder to detect.
What once required a human to place a scam call and respond line by line can now be handled across many conversations at once. That has expanded both the speed and scale of fraud campaigns.
A 44-minute scam call led to Apate’s creation
Apate was founded in 2024 and emerged from research at Macquarie University in Sydney. Founder and CEO Dali Kaafar said the idea took shape after he received a scam call during a family picnic and deliberately kept the caller engaged for 44 minutes. He later decided to hand that task over to AI.
Apate’s agents can pose as victims of different ages, genders, accents, and languages. They answer scam calls, respond to suspicious messages, click phishing links, and even enter scam groups on social platforms. During those exchanges, the agents gather data on bank accounts, crypto wallets, phone numbers, and malicious websites used by scam networks, allowing customers to block the flow of funds before harm occurs.
Early operating results are starting to emerge. Australian telecom company TPG has routed nearly 590,000 calls to Apate’s agents. One banking client used the system to dismantle 22 scam campaigns. Scam losses publicly reported by customer Commonwealth Bank of Australia fell 76%.
On the funding side, Apate completed an $8.15 million seed round in August, bringing total funding to nearly $10 million.
Taiwan’s total fraud losses fell, but phishing-link scams jumped
Taiwan’s fraud picture is moving in two directions at once. Data from the Criminal Investigation Bureau’s 165 anti-fraud dashboard showed 10,652 telecom and internet fraud cases were recorded nationwide in August, with financial losses of NT$4.1 billion. Those figures were down 22.3% and 35.1% from a year earlier.
Phishing-link scams, however, moved sharply higher. Taiwan recorded 505 such cases in August, with losses of NT$54.51 million. Case count rose 99.6% year over year, while losses surged 317.7%. Fake notices involving telecom services and transportation fee collection were the most common forms cited in the report. Even as overall losses declined, fraud groups appeared to be shifting toward methods that are easier to distribute at scale.
Taiwan’s 165 hotline is preparing to adopt AI in stages
Taiwan’s anti-fraud system is also preparing to add AI to its defenses. The Ministry of the Interior said in May that the National Police Agency plans to introduce AI to the 165 hotline in phases from 2027 to 2028. The first stage would automatically convert calls into text, organize case summaries, and provide real-time suggestions to hotline staff. AI voice customer service would be added later in stages.
The same technology is being used in very different ways. Scam groups use it to deepfake public figures and scale up fraud. Apate uses it to imitate victims and drain scammers’ time and resources.