The Numbers Come First
A finance employee at global design firm Arup wired $25.6 million to fraudsters after joining a video call populated by deepfake versions of senior executives. The faces looked real. The voices sounded real. The instructions were false.
That was not an outlier. Starbucks quietly retired an AI inventory system just nine months after deployment after baristas reported miscounts and slower workflows. Deloitte's Australian member firm agreed to partially refund the government for a $290,000 AI-assisted report that contained nonexistent academic sources and a fabricated court quotation.
Three industries. Three technologies. One shared failure: the output could not be trusted.
Trust as Economic Infrastructure
Writing in Fortune as part of a Freedom & Enterprise series marking America's 250th anniversary, Keith Krach — CEO of Freedom 250 and former U.S. Under Secretary of State — frames the problem in terms every CFO and board director should recognize. 'When trust is strong, capital moves, partnerships form, and companies scale,' he writes. 'When it breaks, transactions slow, compliance and insurance costs rise, and leaders retreat from risk.'
Krach defines trust not as blind faith but as 'earned confidence that facts are real, identities are authentic, systems are secure, contracts will be honored, and someone will be accountable when things go wrong.' In business terms, he argues, trust reduces friction. In strategic terms, it creates speed.
The IMF has estimated that severe geoeconomic fragmentation could reduce global output by as much as 7 percent — a figure Krach cites not as an argument for isolation but as a warning about the cost of lost confidence in shared systems.
The Founding Bet and the AI Test
Krach draws a direct line from the Declaration of Independence to the present moment. The signers, he argues, pledged their lives, fortunes, and sacred honor not only to reject a king but to trust citizens they would never meet. That institutional trust became a durable economic advantage: entrepreneurs could start without royal permission, investors could back ideas because property rights and contracts were enforceable, and talent could rise by merit.
Artificial intelligence now tests that same capacity. AI can accelerate science, manufacturing, medicine, logistics, and productivity. It can also clone an executive's voice, generate convincing falsehoods, and automate errors at extraordinary speed. Krach's prescription: 'reliable data, traceable provenance, strong identity controls, independent testing, human accountability, clear rules, and meaningful recourse when systems fail.'
He draws on his own record. At Ariba, he moved procurement from paper onto a global digital network. At DocuSign, he replaced the handwritten signature with a legally enforceable digital agreement. Neither scaled on technology alone — identity, security, and accountability had to be engineered into the product from the start.
The same logic, he writes, applied when he worked to secure global 5G networks as Under Secretary of State. The central question was never who could build fastest or cheapest, but who could be trusted to carry sensitive data across commerce, finance, health care, and energy.
CEO Times Take
Krach is right that the AI competition with China is not a model race — it is a systems race. The ecosystem that earns the deepest institutional confidence will attract the most builders, buyers, partners, and long-term capital. Free-market democracies hold a structural advantage here: accountability mechanisms, independent courts, enforceable contracts, and a press that exposes failure. Those are not soft values. They are hard infrastructure.
The lesson for every board is straightforward: trust is not a compliance checkbox. It is a balance-sheet item. Companies that engineer it into their products and supply chains will move faster and cheaper than those that treat it as an afterthought. The $25.6 million wire transfer was not a technology failure. It was a trust-architecture failure — and the market will keep sending that invoice until leadership treats it as one.
