In-Depth
The AI Trust Problem: Risks and Business Opportunities
- By Pure AI Editors
- 11/03/2026
The phrase "seeing is believing" dates back roughly 500 years. For essentially all of human history, people have relied on their eyes and ears to help determine what is real. A photograph was evidence that an event happened, and a recording of someone's voice was evidence that the person had said something. But artificial intelligence has completely changed this paradigm of reality. AI systems can easily generate deepfakes -- convincing photographs, videos, and audio recordings of events and people that never existed.
The Danger of AI Deepfakes
A bad actor could use AI to clone the voice and create a realistic video of a company's senior financial officer. Then the deepfake would contact an employee in the finance department, and instruct them to transfer millions of dollars to a fraudulent bank account.
A bad actor could create a convincing deepfake video of a political leader appearing to announce a nonexistent military action. The fabricated recording could be distributed widely through social media, during an international crisis. The consequences could be catastrophic.
The AI trust problem extends beyond deliberately deceptive content. AI systems can generate realistic-looking news articles, scientific explanations, and business reports that contain fabricated information. Even when the creator has no intention to deceive anyone, the resulting material can still cause harm if people accept it without checking its accuracy. As AI-generated content becomes more common, distinguishing reliable information from fabricated information becomes increasingly difficult.
AI Deepfake Detection is not Enough
There are many AI detection software systems. Unfortunately, simply using these systems is not a complete answer to a malicious deepfake. Detection tools can sometimes identify AI-generated material, but they can also make mistakes, particularly as generation technology improves. A more reliable approach often involves checking the original source, comparing information with independent sources, and looking for evidence that an image, recording, or claim is authentic. In other words, verifying the provenance of a piece of digital information.
An AI Expert Gives His Comments - Business Opportunities
The Pure AI editors asked AI expert Dr. James McCaffrey to comment. McCaffrey was one of the original members of the Microsoft Deep Learning team -- the group that evolved into Microsoft's current AI efforts. McCaffrey observed, "One of the troubling consequences of the rise of increasingly difficult-to-detect deepfakes is that AI could weaken our overall trust in information."
"When convincing fabrications become commonplace, people will doubt genuine photographs, audio recordings, and news reports. Someone caught making a false statement might even claim that the evidence against them was generated by AI."
McCaffrey added, "A central challenge, therefore, is not simply learning to recognize fake content. It is developing better ways to establish what is true in a world where convincing evidence can be manufactured."
"But the emergence of AI deepfakes provides several new business opportunities. Only the tech giants -- Microsoft, Google, OpenAI, Meta, and so on, have the resources to create a universal digital trust authority. But there is room for small niche services."
"For example, a startup with just two or three people could create a Digital Evidence Authentication Service for insurance claims. The idea is to specialize in verifying whether photographs and videos submitted for vehicle accidents and other claims are genuine or AI-generated. The startup could combine existing forensic software, cryptographic verification, and human review to sell subscription services to small and mid-sized insurance companies without developing its own AI detection technology."