At a technology conference in Singapore this autumn, Interpol's chief information security officer offered a measured but urgent reminder: artificial intelligence is not rewriting the grammar of crime, only accelerating its reach. The ancient arts of deception — impersonation, manipulation, the exploitation of trust — now operate at machine scale, touching thousands of potential victims where once they touched one. The question facing institutions and individuals alike is whether human vigilance can keep pace with a threat that no longer requires a human hand to deliver it.
Interpol warns AI is accelerating cyber fraud, making attacks faster and harder to detect
AI is not inventing new fraud—it is making old fraud devastatingly efficient
Mark
So Interpol is saying AI is making fraud worse. Is that actually new information, or is this just the security establishment catching up to what we already knew?
Mimi
It's more specific than that. Watne is not saying AI creates new fraud methods—he's saying it scales existing ones. The speed and reach are what have changed. One person can now target thousands of victims simultaneously with personalized, convincing messages.
Luke
But that's still somewhat abstract. Do we have examples? Actual cases where AI-accelerated fraud caused measurable damage? The statement is credible, but it's also the kind of warning security officials are supposed to make.
Mimi
Fair point. The source doesn't cite specific incidents. What it does offer is the mechanism: better machine translation, synthetic identities, automation. Those are real capabilities.
Mark
And the advice he gives—identify your critical assets, know your adversaries, tailor defenses—that's not new either, is it?
Mimi
No, it's not. But the framing matters. He's saying don't try to defend against everything. Be strategic. That's actually a useful corrective to the panic response.
Luke
The agentic AI warning is more interesting to me. He's talking about systems that act autonomously. But again, we don't have concrete examples of where this has gone wrong yet.
Mimi
Right. It's a forward-looking concern. Autonomous vehicles, connected devices—the risk is real in principle, but it hasn't materialized into a major incident yet.
Mark
And the trust gap he mentions—people being more cautious with banks than with their smart home devices—that's a behavioral observation, not a technical one.
Mimi
Exactly. It's about how people think about risk, not about the technology itself. The technology is only dangerous if people grant it too much access without understanding what they're allowing.
Luke
So the headline is: Interpol sees AI making fraud faster and harder to detect, and warns about autonomous systems acting without human oversight. Both are plausible. Neither is yet a crisis.
Mimi
That's the right read. It's a warning from someone in a position to see patterns, but it's not based on a specific incident or a quantified surge in attacks.
El Pulso
- AI has not invented new frauds — it has industrialized old ones, allowing a single attacker to run thousands of simultaneous deceptions across languages and borders.
- Machine translation and synthetic digital identities are erasing the telltale signs that once helped people recognize a scam, making the familiar feel dangerously authentic.
- Interpol's security chief urges organizations to abandon the illusion of total defense and instead map their most critical assets, then build protections shaped by specific, known adversaries.
- A new frontier of risk is forming around agentic AI — systems that act autonomously — where errors are no longer just data problems but potential physical events that cannot be reversed.
- The trust gap is widening: users who apply careful skepticism to their banking apps freely grant sweeping permissions to consumer devices, leaving agentic systems to act in spaces no one is watching.
At a technology conference in Singapore this autumn, Interpol's chief information security officer offered a measured but urgent reminder: artificial intelligence is not rewriting the grammar of crime, only accelerating its reach. The ancient arts of deception — impersonation, manipulation, the exploitation of trust — now operate at machine scale, touching thousands of potential victims where once they touched one. The question facing institutions and individuals alike is whether human vigilance can keep pace with a threat that no longer requires a human hand to deliver it.
Speaking at Tech Week Singapore, Interpol's chief information security officer Bjørn R. Watne delivered a clear-eyed assessment: AI is not creating new forms of fraud, but it is transforming their economics. What once required a criminal to craft individual deceptions now unfolds at scale — machine learning handles translation, identity fabrication, and the mimicry of legitimate communication, all at speeds no human operation could match.
The underlying schemes remain familiar. Social engineering has always preyed on the distance between what people want to believe and what they should verify. AI simply widens that distance. A phishing email now reads naturally in any language. A synthetic identity is, on the surface, indistinguishable from a real one. The fraud is old; the delivery is new — and that distinction matters enormously.
Watne's guidance to organizations was deliberately unsentimental. Defending against every conceivable attack is neither possible nor wise. Instead, companies should identify the systems and assets they cannot function without, determine who would want to compromise them, and use threat intelligence to understand how those adversaries actually operate. The strategy looks very different depending on whether the threat is opportunistic or targeted — and misallocating resources is its own vulnerability.
The longer horizon concerns him more. Agentic AI systems — software that acts on a user's behalf without continuous human oversight — introduce risks with physical consequences. A miscalculation in a spreadsheet can be corrected. A miscalculation by an AI governing an autonomous vehicle cannot be undone. And the trust people extend to consumer devices, unlike the caution they apply to banking, tends to be reflexive and unexamined. That combination — autonomous action in a domain where users are not paying attention — is where the next serious harm may quietly take shape.
At Tech Week Singapore this fall, Interpol's chief information security officer laid out a straightforward concern: artificial intelligence is not inventing new ways to defraud people, but it is making the old ways devastatingly more efficient. Bjørn R. Watne told the audience that AI has fundamentally altered the economics of cybercrime. Where a fraudster once had to craft individual deceptions for individual targets, machine learning now allows a single attacker to scale an operation across thousands of potential victims at once. The technology handles the grunt work—translating phishing emails into dozens of languages, generating convincing digital identities, mimicking the texture of legitimate communication—all at machine speed.
The result is a kind of amplification. Criminals have always relied on social engineering, on the gap between what people want to believe and what they should verify. AI simply makes that gap easier to exploit. Better machine translation means a scam email reads naturally in any language. Synthetic digital identities mean the person asking for your password looks, on the surface, indistinguishable from someone real. The fraud schemes themselves are not new. The delivery mechanism is what has changed—and that change is profound.
Watne's advice to companies was practical and unsentimental. Do not try to defend against every possible attack. Instead, identify what actually matters to your business—the systems and assets without which you cannot operate. Then ask who would want to steal or disable those things. Once you know your adversaries, use threat intelligence to understand how they work, what tools they favor, what patterns they follow. Build your defenses around that specific knowledge. The approach differs sharply depending on whether you are dealing with opportunistic criminals testing random targets or sophisticated actors running long-term campaigns against you specifically. Spending resources on protections you do not need is waste.
Beyond the immediate fraud problem, Watne flagged a second concern that is still emerging: agentic AI systems—software capable of taking actions on a user's behalf without constant human intervention. These systems introduce a new category of risk. When an AI makes a mistake in a spreadsheet, the cost is usually recoverable. When an AI system controlling a connected car or an autonomous vehicle makes an error, the consequences can be physical. Someone could be hurt. The damage cannot be undone by pressing undo.
There is also a trust problem baked into how people interact with new technology. In banking, users are conditioned to be cautious. They use PIN codes, they think about security, they hesitate before granting access. With consumer applications and smart devices, that caution often evaporates. People grant permissions readily, without reading what they are actually allowing. They trust the device more than they trust themselves to understand the risk. That gap—between the trust people place in new technology and the actual security posture of that technology—is where agentic AI could do real damage. A system that can act on its own, in a domain where users are not paying attention, is a system that can cause harm before anyone notices it has gone wrong.
Citas Notables
Artificial intelligence is making cyber fraud faster, more scalable and harder to detect, but does not create entirely new criminal methods—it enhances already known deception and fraud schemes— Bjørn R. Watne, Interpol chief information security officer
Quieres la nota completa?
Lee el original en UA.NEWS ↗