Apate AI scam bots now number nearly 197,000 active personas, and the Sydney-based firm has just closed a raise of over $11.3 million to expand those operations into the US and UK, according to The Australian.
The August 2026 Seed round, which PitchBook records at $8.09 million, was led by Silicon Valley VC Lobby Capital, with OIF Ventures, Investible, Concept Ventures, and Baobab Ventures also participating. The Australian’s figure of $11.3 million likely reflects a different round composition or the inclusion of earlier capital; the two figures cannot be reconciled from available sources, so both are reported here.
How Apate AI Scam Bots Actually Work
The core mechanic is straightforward: Apate publishes what it calls ‘dirty’ phone numbers across spam apps and webpages to attract inbound scam traffic, then routes those calls to AI personas trained on hundreds of hours of recorded human scam-baiting sessions. The bots hold convincing phone conversations and engage on WhatsApp and Telegram, each with distinct vocal tics, accents, and the kind of hesitation that makes a mark sound genuine.
Founder Dali Kaafar traces the idea to a 44-minute call he strung along himself at a picnic in Sydney in November 2021. He brought it to doctoral students at Macquarie University, secured funding from the Office of National Intelligence, and spun the project out into Apate in 2023. The Macquarie University research project underpinning the technology built a prototype multilingual conversational AI bot capable of receiving and handling redirected scam calls.
The bots pursue two objectives simultaneously: waste scammer time and extract actionable intelligence. In the six weeks to end-2025, Apate’s bots handled 600,000 scam calls for Australian telco TPG alone, wasting more than 500 days of scammer time and saving an estimated $13 million in potential losses. An earlier phase of the same TPG partnership, documented in a GSMA case study, logged 280,000-plus diverted calls, $7.6 million-plus in prevented losses, 98% scam-traffic detection accuracy, and more than 20,000 impersonated organisations identified, including major banks, government agencies, and telcos.
Crypto Wallet Harvesting and the Intelligence Play
The time-wasting is almost incidental to the intelligence product. In the crypto space, Apate works with what Kaafar describes as one of the leaders in blockchain analysis, extracting new crypto wallet addresses by the hundreds and thousands per engagement cycle. The value is pre-damage visibility: knowing which wallet a scam compound is funnelling funds into before a victim transfers is the entire point.
In July, Apate’s bots uncovered a marketplace where brokers solicited verified bank accounts in India, offering commissions of up to 5% paid in USDT on scam proceeds passing through those accounts. That kind of on-ramp intelligence is precisely what blockchain analysis firms need to trace fund flows before they move further through the chain.
Apate’s partners now include Commonwealth Bank, TPG Telecom, Nvidia, and Microsoft, per the company’s own site, alongside banks across the UK, South Africa, and South East Asia. Startup Daily reported the Commonwealth Bank partnership as a first-of-its-kind deployment in Australia where AI is used not just to detect threats but to actively engage and extract intelligence from scam operations. Australians lost in excess of $2 billion to scams across more than 494,000 reported incidents in 2024, which explains why the banks are writing cheques.
When Scam Bots Meet Scam Bots
Kaafar estimates that 20% to 30% of scam text conversations already deploy AI on the attacker side, and he is not particularly concerned about parity. His game-theory argument: a defensive bot trying to extract intelligence has a structural advantage over an offensive bot trying to trigger an action, because extracting information from an AI system is easier than manipulating one into performing a specific behaviour.
‘You can also demonstrate mathematically that that is to the advantage of a defender because it becomes easier to extract intelligence from the attacker’s AI model,’ he says.
Global scam operations generate an estimated $1.24 trillion annually, so the compute budget on the offensive side is not the constraint. Apate’s bet is that scale and intelligence depth are. One of its internal KPIs is the monthly count of expletives frustrated scammers direct at its bots: a proxy for how convincingly the personas are holding their ground.
With the latest capital deployed toward US and UK hiring, the key question is whether Apate’s TPG-style partnership model translates to markets where telco regulation and data-sharing norms look very different from Australia’s. The UK rollout will be the first real test.
