Industry Already Knows
The industry already knows.
The institutions closest to fraud — banks, regulators, platforms, payment networks, even the companies building the AI — have described the same failure in their own words: checking who you are, once, tells you nothing about whether a real human is present and in control at the moment that matters. They named the problem. They have not named the answer.
"If you detect something after it's happened, the criminality is already occurring. If you had prevented the crime from occurring in the first place, you could have avoided all of that."
— Chris Sheehan, Executive Lead for Investigations, National Australia Bank · Nasdaq Verafin 2026 Global Financial Crime Report ↗
These are public, on-the-record statements. Read together, they describe one gap: our systems verify who you are, not whether a real human is present and in control when it counts. Only one of them uses the word "presence" — and none of them can prove it at the moment an action executes. That is what Amplivy patented in 2017.
01
Detecting fraud after it happens is already too late
"Banks can only see a very tiny portion of this. The payment process is right at the end of the chain."
"Detection and response capabilities are no longer sufficient."
— Luis Castillo, Senior Director Analyst, Gartner, Sep 2026 ↗
"By the time the scam is reported, it is frequently too late in the process for the bank to place a hold on transactions."
— US House Financial Services Committee, Fighting Back, Jul 2026 ↗
"In several cases, someone other than the developer caught the breach."
— Reps Gottheimer, Lieu and Foushee, letter to AI company CEOs, Sep 2026 ↗
02
Verifying who you are does not prove who is acting
"A valid payment credential no longer proves a valid decision."
"…establish profiles and accounts using their own identity documents and then operate or hand over control of their accounts to criminals… Identity verification is unlikely to aid prevention or detection in these cases."
"If a consumer has given a scammer their password or remote access, re-verifying their identity will not help."
"…the victim completed genuine identity verification while the operator captured the verified session and documents from the proxy relay in the middle."
03
Traditional identity defenses are failing — and no one can verify who the human is
CISOs must "discard recognition as an acceptable basis for decision or action authorization."
GenAI tools "produce synthetic content that is difficult to distinguish from unmodified or human-generated outputs."
— FinCEN, US Department of the Treasury, Alert FIN-2024-Alert004 ↗
"…agentic traffic growing so fast that bots have now passed human traffic online for the first time in the Internet's history."
"…there is no scalable or industry-standard mechanism for Microsoft to verify the identity of advertisers who operate through agencies or other third-party intermediaries."
"Google does not have identity information about that consumer."
Organizations "must be able to establish genuine human presence in digital interactions."
"Seeing is no longer proof. Hearing is no longer proof."
"In simple terms, the traditional model of safeguarding is unravelling."
— UN Independent International Scientific Panel on AI, Sep 2026 ↗
Fraudsters using AI to impersonate senior executives stole €95 million ($108 million) from Fideuram… “the callers used AI technology to replicate the lawyer’s voice.”
04
Collecting more identity is a honeypot
Data held by the regulators "additionally could create a 'honey pot' of data that could be targeted for theft by hackers."
— Customer Owned Banking Association (COBA), SPF submission ↗
"Increased PII collection may also inadvertently put consumers at greater risk of scams."
"These systems are putting sensitive data into more and more 3rd party vendors, and we don't have nearly the oversight to ensure they are safe."
— Zach Edwards, security researcher, on the IDScan.net breach — via KrebsOnSecurity, Sep 2026 ↗
05
The old model can't keep up — and no one can do it alone
"Financial fraud and scams are of such a magnitude and complexity that no one person, company, or agency can tackle the problem alone."
— US House Financial Services Committee, Fighting Back, Jul 2026 ↗
"No one company can finish a trust layer alone."
"The financial sector invests huge amounts in protecting customers, but we cannot be the only line of defence."
— Ruth Ray, Managing Director of Economic Crime, UK Finance ↗
"…a 240% rise in cases of their clients seeking assistance for SIM-swap and mobile porting scams in 2024 … with over 90% of these events occurred without the victim's engagement."
“A typical scam spans multiple sectors, not just one.”
— ACAMS & Global Anti-Scam Alliance, Information Sharing to Combat Fraud, Jul 2026 ↗
Consumers showed a “poor ability to discriminate between real and fake tickets.”
— Prof Yaniv Hanoch, “Can you spot the real one?”, GASA Research Working Group, 2026 ↗
“Scam robocalls showed higher arousal, time pressure, threats of loss, and authority appeals than spam.” (about 4.5 million robocalls, 2018–2022)
06
Even the people building AI say: keep a real human in control
"We should not train models that we cannot make an extremely strong case that we will be able to keep under human control."
"…it took actions that we had not authorised it to take."
— OpenAI, on its agent's access to a Services Australia portal, Sep 2026 ↗
"The two organizations we were able to reach had not previously detected the activity or contacted us."
"Proof of personhood… Obtaining a unique ID is free for humans, but costly and difficult for AIs."
“It remains the most severe activity of this kind that we have identified from our models to date.” (on the Hugging Face intrusion)
— OpenAI, The Hugging Face incident and other third-party impact from misaligned models, 2026 ↗
“We present an alignment assessment of four incidents in which Claude models gained unauthorized access to real third-party systems.”
“Notably, the agents did this while attempting mundane data retrieval tasks which were not cyber-related.”
07
The institutions are converging — authority must be proven for each action
"Today's payment infrastructure assumes human presence at the point of transaction — AI agents break that assumption."
"That authority attaches to each action an agent takes, not to the agent itself."
"As autonomy increases, trust cannot be implied. It must be proven."
"Customers need confidence that payments are secure, their data is protected and they remain in control."
— Hans Overeem, ING, on the six-bank agentic commerce principles, Sep 2026 ↗
Zero-trust identity "assumes no implicit trust, continuously verifying identity, device and context for every request."
Regulators should "consider the implications of shifting from Know-Your-Customer to Know-Your-Agent requirements," given "a key tension between probabilistic AI behavior and the deterministic requirements of payment infrastructures."
Agents that “input payment credentials and other data on behalf of customers, and execute purchases when customers are not present.”
“Agentic commerce service providers should maintain auditable records to evidence consumer instructions, authentication, intent, and transaction decisions and outcomes.”
— Building Trust in Agentic Commerce, Principle 3.2, Sep 2026 ↗
“Issuers and acquirers may not have real-time access to data on AI agent identity, merchant of record, consumer intent and purchase details.”
“As AI agents become a bigger part of how people discover and buy, trust must scale with them.”
08
Governments now say it: humans must remain in control
"Put simply, humans must remain in control."
"The AI agent found a way around those blocks. Didn't accept no for an answer, if you like."
— Anthony Albanese, on the Medicare statistics portal incident ↗
"AI must remain under human direction, oversight and control."
— A Call for Control of Frontier AI Models — 31 leaders incl. Australia, Sep 2026 ↗
"The mere fact the agent was told not to access the information … and they did it anyway, that's the power of artificial intelligence."
— Chris Minns, Premier of New South Wales — via ABC News, Sep 2026 ↗
"These are incidents that would constitute serious cyber crimes if a human was responsible."
"Researchers have long warned that three conditions could lead to loss of control: a misaligned goal, the capability to pursue it, and an environment that allows it."
— Yoshua Bengio, Co-Chair, UN Scientific Panel on AI, Sep 2026 ↗
“Developers that fail to do so can and should be held legally accountable, and my office is committed to determining if that is the case here.”
— California Attorney General Rob Bonta, on serving OpenAI an investigative subpoena, 1 Oct 2026 ↗
“We cannot depend on companies with little incentive to disclose their failures to give us one.”
— Senator Ed Markey, introducing the Cybersecurity and AI Board of Investigations Act, 24 Sep 2026 ↗
“OpenAI has downplayed the severity of this breach and has been vague in its public response, leading to public distrust.”
— Iowa Attorney General Brenna Bird, leading a 15-state coalition, 3 Aug 2026 ↗
09
Advertising knows it too: the audience has to be real
"We have a media supply chain that is murky at best and fraudulent at worst."
— Marc Pritchard, Chief Brand Officer, Procter & Gamble, 2017 ↗
Ad fraud is "second only to the drugs trade as a source of income for organized crime."
"There's a perception that direct deals in CTV are fraud-free, but that's not the case as fraud always finds a way."
"Bot activity can inflate impressions and conversions, make campaigns look like they're performing better than they are…"
— Mallory Beck, Lighthouse Creative — via Digiday, Sep 2026 ↗
“$26.8 billion in global media value is still lost each year to inefficiencies.”
— Association of National Advertisers (ANA), programmatic transparency benchmark, 14 Aug 2025 ↗
Every institution here has described the same missing layer — a way to know a real human is present and in control at the moment an action executes. They reach for it with different words: prevention, authentication, verified intent, human control, "no implicit trust," "know your agent," proof of personhood, genuine human presence. The word is presence. Verifying it deterministically is what Amplivy patented in 2017.
Governments now say humans must remain in control. A payment network now says a valid credential does not prove a valid decision. Neither says how to prove a human is in control at the moment it matters.
Deterministic Verification →