When the Assumptions Fail

An email will send. A meeting will connect. The information we receive will be accurate. The people responsible for protecting us will act when they recognize a threat.

We organize our lives around assumptions like these.

But what happens when those assumptions are no longer true?

Recent headlines give us three reasons to take that question seriously—and to examine whether our preparation has kept pace with our dependence.

1. When the system is unavailable

The widespread Microsoft outage affecting Outlook, Teams, and other services has exposed how much ordinary life depends on digital infrastructure. Associated Press reporting described users unable to send or receive email and disruptions across Microsoft’s applications.¹

A message cannot be delivered. A team cannot coordinate. Information needed to serve someone is suddenly out of reach.

The outage does not establish that AI caused the disruption. It does remind us that services we increasingly depend on can become unavailable.

As we integrate AI into more of our work, we need to understand what supports it: electricity, networks, data centers, chips, vendors, and people.

The application on the screen is only the part we see.

For a church, an interruption can complicate reaching members. For an organization, it can interfere with essential coordination. For a household, it can block access to information needed to make a timely decision.

Readiness begins with understanding those dependencies before they are interrupted.

2. When the system is wrong

In September, CNN reported that faulty AI-assisted intelligence had nearly prompted a U.S. military operation against a Chinese vessel.

According to CNN’s sources, an analyst used a chatbot to help assess intelligence concerning the ship’s cargo. The resulting report incorrectly identified the cargo as components for a nuclear weapons program.

Military preparations were underway before officials examined the report more closely and discovered the mistake. One source told CNN the episode “almost started a war.”²

That description belongs to the source. The reporting describes preparations for an operation that could have escalated into conflict, rather than an established decision to declare war.

Further scrutiny caught the error before the planned operation proceeded.

That distinction matters. So does the lesson.

A polished report can carry a false conclusion. A confident answer can still require verification.

Human review must mean more than having someone available to approve what a machine produces. It requires people with the knowledge, time, and authority to question it.

Sometimes the most important person in the room is the one willing to say, “Before we act, let’s verify this.”

3. When the people and incentives require scrutiny

The scandal surrounding Nvidia and its now-former chief executive, Jensen Huang, raises questions about another kind of vulnerability.

Nvidia supplies chips used to develop and operate AI systems. Reporting about Huang’s relationship with a businesswoman alleged to have ties to Chinese intelligence has raised concerns about access, sensitive technology, and national security.

Precision is essential here. CNN has reported disagreement among officials about the evidence. The espionage allegations have not been conclusively established.³

Unresolved allegations still require a responsible response: investigate credible risks, establish the facts, and explain the decisions.

The controversy also extends to export enforcement. Reuters reported that Treasury Secretary Scott Bessent pointed to the Commerce Department over the failure to impose restrictions on Nvidia customer Megaspeed.⁴

The public deserves to understand what warnings officials received, how they evaluated those warnings, and whether safeguards worked as intended.

Then there is the financial arrangement.

The administration imposed a 25% tariff on certain advanced computing chips, including Nvidia’s H200, with exemptions for specified domestic uses. The White House described the measure as part of its economic and national security policy.⁵

Democratic lawmakers Ron Wyden and Richard Neal have opened an investigation into the arrangement allowing certain chips to enter the United States for review before receiving authorization for export to China. Their announcement challenges both the legal mechanism and its national security consequences. Those are the lawmakers’ allegations and arguments, rather than settled findings.⁶

This is federal tariff revenue—not a blanket 25% share of Nvidia’s profits, and not a payment to Trump personally.

Even with that distinction, a serious governance question remains:

When the government receives revenue from transactions it must also evaluate for national security risks, what safeguards keep financial considerations from compromising its judgment?

Receiving tariff revenue does not, by itself, prove corruption. Nor does it establish why any particular official acted or failed to act.

It does make transparency important.

A national security decision must remain defensible even when the safer choice brings in less money.

4. What accountability should produce

Democrats are preparing to take control of Congress in January, while Trump remains president. That creates a different oversight environment. Democratic lawmakers have already demanded records concerning the Nvidia arrangement.^{6, 7}

My concern is what comes from that scrutiny.

We need investigations that establish facts, rules that address real vulnerabilities, and accountability that improves public protection.

Responsible oversight can help innovation earn the trust its widespread adoption requires.

Speed matters. So do accuracy, security, reliability, and the ability to recover when something goes wrong.

The future of AI will be shaped by decisions about access to advanced chips, export controls, procurement, verification, and responsibility for consequential mistakes.

We should examine those decisions by asking whose interests they serve and how the public can assess the results.

5. What readiness looks like for us

These questions reach beyond Washington and Silicon Valley.

For a church, readiness means another way to reach members when email is unavailable.

For an organization, it means people who can continue essential work when an automated process stops.

For a household, it means access to critical information when an account, device, or connection fails.

It also means checking whether our backups are independent. Two applications offer limited protection if both depend on the same unavailable service.

Before we depend on a system, we should ask:

• Availability: What happens if we cannot access it?
• Accuracy: Who checks whether its answers are correct?
• Authority: Who can stop a harmful decision?
• Continuity: What other way do we have to keep serving people?

A second tool is useful only if it provides a workable alternative. Another person’s review is useful only if that person can independently examine the evidence.

Backups, independent verification, and clear responsibility address different vulnerabilities. We need to plan for each.

6. Preparation is an expression of care

As a pastor, I hear the wisdom of Proverbs 27:12 in this moment: the prudent recognize danger and take precautions.

Preparation is an expression of care for the people entrusted to us.

We do not have to predict every disruption to identify essential responsibilities. We can decide what must continue, who will carry it forward, and what they will need if the usual system is unavailable or unreliable.

That work belongs in our planning now.

Next week, I will release a report taking this AI readiness conversation further. I am also highlighting updates and new material on this website to help households, churches, and organizations turn concern into practical action.

Explore the readiness resources, examine the reporting behind these issues, and ask your own questions.

We cannot control every outage, corporate decision, or government failure.

We can examine our dependencies. We can verify before we act. We can prepare another way to keep serving.

When the system stops—or gives us the wrong answer—the people depending on us still need us.

Build it before you need it.

• AI Readiness
• Leadership
• Accountability
• Technology
• National Security