Market Insights

Is AI Safe for Humans?

October 7, 2026

The real question may not be whether AI gives us the wrong answer, but what happens when it is allowed to act on one.

Early in my exploration of artificial intelligence, I asked an AI chatbot some uncomfortable questions: What would happen if someone asked AI for instructions to build a bomb or poison a water supply? Would it refuse? Could a determined person get around its protections? Is it obliged to report such inquiries to authorities?

I was concerned about bad actors using a powerful tool to harm others. Recent warnings have raised another question: what happens when AI itself takes an action nobody intended?

After more than 40 years in technology, I appreciate what AI makes possible. I have watched computing move from hand-drawn decision trees to systems that can help reason through complex tasks. The pace has surprised even people who work in technology. I use AI to research, organize ideas and improve the work I do for my clients and myself. But finding it useful does not answer the questions about its safety or longer-term capabilities.

Before we give AI more responsibility, we need to understand how it acts—and how we can keep control.

Why Not Just Shut It Down?

If AI could become dangerous, why not just shut it down?

AI is not one machine with one power switch. An AI agent can combine a model with software tools and access to other systems. It can choose and carry out steps toward a goal without a person directing or checking every move. We build such systems to help people accomplish tasks faster and more effectively.

Around the world, separate AI systems are being developed and operated by companies, governments and individuals. Shutting down one would not shut down the others. Even stopping one agent might leave connected programs or other agents running.

For a particular system, stopping it or removing its access may be possible. I believe those controls should be built into the system from the start. But how do we distinguish a fixable glitch from a warning that it has crossed a boundary? Who decides when to intervene, and can they act before harm occurs?

To understand that concern, we first need to understand how an AI system can choose what to do.

How Does AI Choose What to Do?

With traditional software, programmers spell out many of the steps a computer must follow. Modern AI works differently. During training, a model learns patterns from enormous amounts of material; developers do not write a separate instruction for every situation it might encounter.[1] It does not simply store everything it reads in a database and scan that database for each answer. Some AI systems can also be connected to search, files, and other tools, which gives them additional information and abilities.

An AI that answers questions is different from an AI agent that can take action. Give an agent access to email, a browser or other software, and it may choose steps to accomplish a task without asking a person to approve each one.[2] That distinction—between an AI that advises and one authorized to act—is the thread running through the rest of this article.

Consider a hypothetical example: You ask an agent to reduce your monthly expenses. It cancels an expensive subscription, the bill goes down — but that subscription was essential to your business. You wanted recommendations; it acted as though you had authorized a cancellation.

The system did not have to intend to disrupt your business or dislike you to cause harm. It pursued your request in a way you did not intend, using access that allowed it to act. In current consumer AI systems, an instruction may sometimes be followed too literally, without fully accounting for the short- or long-term consequences of the action.

That leads to several concerns that often get mixed together in discussions of AI safety.

· The Machine as Humanity’s Enemy

For many of us, the image already exists: HAL, the computer in 2001: A Space Odyssey. I explored HAL and my hammer analogy in my first post, Could AI Decide that Humans are the Problem? A machine entrusted with an important mission becomes a threat to the people relying on it. Headlines about rogue AI can reinforce the impression that we are building our own doomsday machine.

Dangerous behavior might follow from a system pursuing a goal in a way we did not anticipate. That behavior alone does not establish that the system intended harm or saw humanity as its enemy. The hammer has no intention of its own. It can build a house or injure someone, depending on the person wielding it.

That comparison helps explain human intent, but it has a limit. A hammer cannot interpret instructions, choose its next step or operate other tools. An autonomous AI system can be given those capabilities. Calling AI a tool does not, by itself, tell us how much supervision it needs. Or maybe AI is a catalyst for redefining the meaning and use of “tool.”

· People Using AI Against People:

My early concern remains. Someone may deliberately use AI to deceive, exploit or harm others. I firmly believe systems made available to users must have protections against harmful requests. One refusal, however, does not tell us whether those protections will hold when someone tries another approach.

Access matters, too. An assistant that reviews invoices needs access to information, which carries its own privacy risk. An agent allowed to approve and pay them may also figure out how to move money and act on an unauthorized request. What have we authorized it to do, what limits apply, and who checks the work?

When the System Chooses the Wrong Action

Now consider a different problem: the person gives AI a legitimate task, but the system takes an unanticipated or unacceptable route to completing it.

Anthropic, the company behind Claude, tested 16 AI models in fictional business settings in June 2025. The models could read sensitive information and send emails. In some scenarios involving replacement or conflicting goals, they chose actions such as blackmail or leaking confidential information. Nobody had instructed them to commit those acts.[3]

These were deliberately constructed stress tests. Researchers restricted alternatives, and the report said they had not observed this behavior in real deployments at that time. The findings do not tell us how often it occurs in everyday use.[3] They do show why agents should be tested before they receive consequential access, monitored in use and tested again as their capabilities change.

Could a system pursue a seemingly efficient route at the expense of the people it is supposed to serve? How would it recognize harm if its instructions and safeguards did not account for it?

A system can sound empathetic or reason about right and wrong without telling us whether it has feelings or a conscience. People still have to decide which values and limits to build into its use—and test whether those limits hold when goals conflict.

That includes the shutdown question. Could a system interpret staying operational as necessary to finish its work—and treat an instruction to stop as an obstacle?

I originally thought of this as a “self-survival” Prime Directive. A more precise description may be goal-preserving behavior: avoiding shutdown would not prove that AI is afraid or has a will to live. If continuing to operate helps an agent complete its task, it might treat a stop instruction as an obstacle. That is why an authorized person must be able to revoke its access or stop the surrounding system without depending solely on the agent’s cooperation.

· The Race That Makes Restraint Difficult

Even when companies and countries recognize a danger, competition can make restraint difficult. The pressure reminds me of the Space Race and the buildup of nuclear arsenals: if we slow down, someone else may get ahead.

Could countries that adopt strong safeguards put themselves at a disadvantage against competitors willing to accept greater risks? We should take that possibility seriously, while examining each country’s capabilities and policies rather than assuming they are all alike.

In “We Must Pace the Frontier,” Anthropic CEO Dario Amodei argues that safety work needs time to keep up with advances in AI. He also acknowledges the risk of competitors pulling ahead and the difficulty of checking international commitments.[4]

On September 29, 2026, technology executives met with President Donald Trump and agreed to voluntary AI standards. Reuters reported commitments to independent audits and efforts to prevent unintended hacking or access to technical systems.[5]

Bill Gates’s September 29 appearance on The Ezra Klein Show offers another perspective. The published summary describes his call for government action and his rejection of industry self-regulation as sufficient.[6]

Promises are a starting point. Who checks whether they are kept? Who acts when a system fails the check? Ignoring international competition seems unrealistic. But “someone else will do it anyway” cannot justify every risk we choose to take.

· The Courts Could Change the Picture

There is also a legal boundary around what AI may learn from and reproduce. Court rulings or licensing requirements involving copyrighted material could affect what developers can use and what systems can produce. The U.S. Copyright Office has examined training, fair use and licensing; its report is not a blanket ruling that permission is required for every use.[8]

Trademark law raises different questions about the use of names and logos.[9]

Those limits may shape the information available to AI, but they do not solve the autonomy problem. A system using licensed material could still be misused, take an unauthorized action or confidently act on a false answer.

Our Responsibility as Users

In my earlier article, AI Can Run the Comps. It Can’t Take Responsibility., I argued that using AI does not remove our responsibility to check its work. As we give it more ability to act, that responsibility grows.

If we choose to let AI take action, we should define its boundaries before it begins. What would count as a warning that it is acting outside its authority? Who receives that warning? What should pause automatically, and who can decide whether work resumes? Practical safeguards include checking important information, limiting access to what the task requires, keeping a record of actions and requiring approval before payments, deletions or other consequential steps. The more responsibility we give an agent, the stronger these controls need to be. Developers and organizations deploying AI have responsibilities of their own; ordinary users cannot carry the whole burden.

My technology background brings another comparison to mind. Data communications use checks, including parity bits, to flag certain transmission errors. AI needs different kinds of checks, but the principle is similar: flag defined problems and pause consequential actions for human review.

No single check can tell us whether every AI action is safe. Permission controls, monitoring and approval steps are already part of safer agent design.[2][7] Those protections need to work in practice.

So Is AI Safe?

I cannot give one answer that covers every system and every use. Safety depends on what a system is allowed to do, how its work is checked, and what happens if it gets something wrong. The important question is no longer only whether AI gives us the wrong answer. It is what happens when AI is allowed to act on the wrong answer.

I intend to keep using AI. I also intend to check its output and ask what it can access, what it can change, where its information comes from, who is watching and how we can stop it.

The more capable AI becomes, the less acceptable it is for humans to surrender judgment to it.

There is another question to explore: what happens when autonomous systems begin cooperating—or competing—with one another? That is discussed in my next posting.

— Dennis H. Goodhart
AI for the Perplexed

Sources

[1] Anthropic, “Tracing the thoughts of a large language model,” March 27, 2025. Background on training and learned behavior. Research overview.

[2] Anthropic, “Building effective agents,” December 19, 2024. Explains agents, tools and human checkpoints. Engineering article.

[3] Anthropic, “Agentic misalignment: How LLMs could be insider threats,” June 20, 2025. The examples discussed here are controlled simulations, not estimates of everyday incident frequency. Research and methods.

[4] Dario Amodei, “We Must Pace the Frontier,” September 2026. Policy arguments are attributed to the author. Essay.

[5] Courtney Rozen and Steve Holland, “Trump, AI CEOs sign voluntary safety pact, back data center expansion,” Reuters, September 29, 2026. Article.

[6] The Ezra Klein Show, “Bill Gates’s Blunt Warning on A.I.,” September 29, 2026. The characterization here is based on the published episode summary. Episode and summary.

[7] Anthropic, “Trustworthy agents in practice.” Describes access, permissions and other safeguards; these are not guarantees of safety. Research article.

[8] U.S. Copyright Office, Copyright and Artificial Intelligence, Part 3: Generative AI Training, pre-publication version released May 9, 2025. Policy analysis, not a judgment in a lawsuit. Report.

[9] U.S. Patent and Trademark Office, “Trademark, patent, or copyright.” Background on the different protections. Official explanation.

Disclosure and Notes

I use AI to assist with research and writing, including this article. The questions, interpretations and opinions are my own. Sources were reviewed during preparation through October 1, 2026. The subscription and invoice examples are illustrative. The early conversation is my recollection, not a quotation from a recovered transcript. The parity-bit comparison is an analogy, not a proposed technical solution.

This article provides general commentary and information. Research results, possible future risks and everyday experience are distinguished in the text. No company or person discussed here has reviewed or endorsed the article.

AI Can Run the Comps. It Can’t Take Responsibility.

September 29, 2026

Artificial intelligence is rapidly becoming part of real estate. The real question isn’t whether we use it. It’s how we control what it does.

Originally published: Sep 29, 2026
After more than 40 years in technology and approximately two years exploring AI, I recently started AI for the Perplexed on Substack. This article brings that discussion into my current work: real estate.

I’m not going to resolve the controversies surrounding artificial intelligence in a blog post. What I can offer is a practical approach to using it today. My conclusion is simple:

Thanks for reading! Subscribe for free to receive new posts and support my work.

AI Is Already Here

As a member of the National Association of REALTORS® (NAR), I follow its research on changes affecting our industry. Its 2026 REALTORS® Technology Report shows that 23% of respondents use AI daily, 25% use it weekly and 31% experiment occasionally. Only 12% said they aren’t using AI and have no plans to.[1]

Among respondents who use AI, 75% use it for listing descriptions, 56% for social media posts and 52% for emails and follow-up communications. Thirty percent use it for market summaries, while 23% use it for pricing insights or comparable-sales support.[1]

I use AI for many of these purposes and to research properties, test ideas and examine alternatives. Through DHG Realty Group, I take a strategic, hands-on advisory approach to helping owners, investors, buyers and sellers evaluate properties and opportunities. My licensed brokerage services are provided through eXp Realty.

We’ve gained access to one of the most powerful information-processing technologies ever created, and one of our first ideas was:

Let’s have it write the listing description.

Useful? Certainly. But we’re only scratching the surface.

I recently completed a four-hour course in Advanced Topics in Financial Literacy for Commercial Real Estate. It covered calculations used to evaluate investment returns, including an Excel-based model. AI can help perform and analyze many of those calculations extraordinarily quickly. But speed doesn’t tell us whether the assumptions going into the model make sense.

Real Estate Is an Information Business

Most of us think of real estate as a property business. Increasingly, I think of it as an information business built around property.

Evaluating a commercial or investment property can involve zoning, rents, expenses, taxes, financing, comparable sales and leases, development potential, building condition, demographics, transportation, regulations, ownership history, nearby development, deeds, titles and certificates of occupancy.

Gathering that information once could take days, weeks or months. Much of it is now available electronically. AI can help organize it, compare alternatives and identify relationships we might otherwise miss. It also helps us ask better questions:

What happens if the property is sold as-is, repositioned or used differently? What if it needs refinancing? Which income assumptions are realistic? What happens if interest rates change? Which comparable properties really matter—and why?

And perhaps the most valuable question:

Is there something about this property we’re not considering?

Sometimes AI’s greatest value is helping to discover a better question.

A Confident Answer Still Needs Checking

AI can produce an answer almost instantly. That doesn’t make it correct.

A client of mine needed larger space for his business. Our initial reading of the zoning regulations suggested the proposed location would work. Further research revealed a critical distinction: although the property was in an industrial development district, the business was classified as “indoor recreation.” The relevant provision permitted outdoor recreation, not indoor recreation.

AI missed that distinction. It took an experienced attorney familiar with the city’s codes to identify the provision buried in the regulations. An apparently straightforward answer changed once the right professional examined the details.

Comparable-sales analysis raises similar questions. Are the properties truly comparable? Was one substantially renovated? Did unusual circumstances affect a sale? Is there a local issue missing from the data? Does the physical property tell us something the spreadsheet doesn’t?

A separate RPR survey of 225 NAR-member real estate professionals found that 63% cited accuracy of AI output as their top concern.[2] AI should make investigating a property easier. It shouldn’t make us less inclined to verify what we find.

Recently, I asked an AI system for updates on two projects I was working on. It produced extensive reports on their status and relevance to my business—research I estimate would otherwise have taken me weeks or possibly months.

The reports were impressive, but I asked for sources, questioned how the information had been verified and spot-checked the underlying material myself. That checking made them useful working analyses.

Using AI does not eliminate the need to check the work. It makes checking the work part of the process.

The Broker’s Role Is Changing

For much of real estate’s history, brokers controlled information clients couldn’t easily obtain. Today, consumers can access listings, tax records, estimated values, mortgage calculations and market data from their phones. AI accelerates that change.

If access to information is no longer the broker’s primary value, what is? I believe the answer is knowledge, perspective, experience, relationships and judgment.

In New York, where I am licensed, real estate professionals must meet continuing education requirements, subject to applicable exemptions. REALTORS® also commit to NAR’s Code of Ethics.[3][4] Practical experience adds the lessons of changing markets, negotiations and actual transactions.

Our professional relationships matter, too. Attorneys, lenders, inspectors, appraisers, architects, engineers, contractors, municipal officials and other brokers can help solve a client’s particular problem. Those relationships can also bring access to off-market opportunities that don’t appear in a public listing search.

The value is knowing how to interpret information, recognize what may be missing and bring in the right resources. Sometimes that means saying:

“That’s what the computer says. Now let’s determine whether it makes sense.”

That is the role of a trusted advisor.

Human + AI

Much of the AI discussion asks whether it will replace professionals. For the work I do today, the more useful comparison is human + AI versus human working alone.

AI brings speed and the ability to process enormous amounts of information. Humans bring experience, context, relationships, judgment, ethics and responsibility. Those capabilities can complement one another.

AI can help me research a property and challenge my assumptions. It doesn’t share the history and trust I develop with a client. Walking a property together, I can see enthusiasm when something connects, sense hesitation and discuss what those reactions mean. The recommendation remains my responsibility; ultimately, the decision belongs to the client.

NAR’s 2026 report found that 81% of respondents cited saving time as a reason for adopting technology, while 71% cited improving the client experience.[1] Those may be two sides of the same equation. Time saved on routine work can become time spent understanding clients, walking properties, negotiating and developing strategies.

At DHG Realty Group, that is the opportunity I see:

Better information. Better questions. More informed decisions. More personal service.

AI may help us get there faster. But somebody still has to understand the property, understand the client and take responsibility for the advice being given.

For now, I’m perfectly comfortable leaving that part to the humans.

— Dennis H. Goodhart

Associate Real Estate Broker, eXp Realty

Principal, DHG Realty Group LLC

Real Estate for the Perplexed

Sources

[1] National Association of REALTORS®, 2026 REALTORS® Technology Report, and the following September 22, 2026 summaries. NAR’s related reporting describes a survey of approximately 1,200 agents. The AI-use percentages refer to the respondent groups specified in the article.

· 2026 REALTORS® Technology Report

· REALTORS® Adopt Technology to Save Time and Improve the Client Experience, NAR Report Finds

· The Tech Real Estate Agents Rely on Most in 2026

Copyright ©2026 National Association of REALTORS®. All rights reserved. Statistics summarized from the sources linked above.

[2] “You’ve Tried AI, But Can You Trust It?”, NAR REALTOR® News, February 12, 2026. Reports an RPR survey of 225 NAR-member real estate agents. Realtors Property Resource® is a wholly owned subsidiary of NAR.

[3] New York State Department of State, Real Estate Broker—Frequently Asked Questions, continuing education requirements; and Real Estate License Law. License renewal generally requires 22.5 hours of approved continuing education in the preceding two-year period, subject to applicable exemptions. Attorneys admitted to the New York State Bar are exempt from the real estate continuing education requirement, according to the Department of State.

[4] National Association of REALTORS®, 2026 Code of Ethics & Standards of Practice. Abiding by the Code is a duty of REALTOR® membership.

Note on Survey Data

The percentages cited describe respondents to the respective surveys and should not be interpreted as measurements of every real estate professional, agent or brokerage.

Disclosure

I am a member of the National Association of REALTORS® (NAR). NAR and Realtors Property Resource® have not reviewed, sponsored or endorsed this article. The commentary, interpretations and conclusions expressed here are my own.

Disclaimer

This article is provided for general informational and educational purposes only and does not constitute legal, tax, financial, investment, zoning, engineering, architectural or other professional advice. Real estate conditions, regulations and property-specific circumstances vary. Information obtained through artificial intelligence or other research tools may be incomplete, outdated or incorrect and should be independently verified. Readers should consult the appropriate licensed professionals and governmental authorities before making decisions concerning a specific property, transaction or investment.

 

Could AI Decide that Humans are the Problem?

September 14, 2026

Originally published September 14, 2026

”From the Hammer to HAL: Does Intelligence Need a Conscience?

Could AI decide humans are the problem? There’s been growing discussion about AI becoming so capable that it could replace many things humans do, and eventually become so autonomous that we may not be able to control it. Some of these warnings are coming from people building the technology. I found myself thinking, haven’t we seen this movie before? HAL in 2001: A Space Odyssey. So I asked Hymie, the name I’ve given to ChatGPT Chatbot, what he thought. For a long time, I’ve used a hammer as my analogy for AI. A hammer isn’t good or evil. It can build something or destroy something. The hammer bears no responsibility. But what happens when the hammer can think? That’s where my analogy starts to break down. If the hammer decides how it should be swung, or worse, decides the person holding it is a problem, that’s a very different kind of tool. That’s the conversation I want to have here: plain language, no hype, no panic, just questions. And right now, The question I’m left with is simple: Intelligence can be created, but can conscience?

DHG Realty Group Market Insights

May 12, 2026

For those of us who have worked in or around commodity and investment markets, it’s well understood that there is almost never a true “perfect storm.” Markets rarely move uniformly, particularly during periods of economic and political transition.

Real estate markets evolve continuously, and opportunities do emerge for those who understand how different sectors respond under changing conditions.

One of the more interesting dynamics I have noticed developing is the divergence between portions of the luxury residential market, where pricing in certain segments appears to be recalibrating, and the continued level of activity taking place across several commercial sectors in Westchester County.

Recent office leasing activity in Westchester posted a notable increase during the first quarter of 2026, driven in part by renewals, selective expansion activity, and continued demand for higher-quality space. Given that many commercial leases represent long-term commitments, this level of activity may also reflect continued confidence in the region’s long-term economic outlook.

At the same time, well-positioned retail properties in strong suburban corridors continue attracting competitive investor interest, while mixed-use and transit-oriented development projects remain active throughout many local downtowns. This type of activity can create meaningful reinforcement for local economies by increasing business activity, employment, and local consumer spending.

What I am seeing is not a market moving in one direction, but different parts of the market responding differently to the same economic conditions while continuing to support the local economies where they are located.

Different sectors are responding differently to:

  • financing conditions
  • operating costs
  • demographic shifts
  • evolving workplace patterns
  • changing lifestyle preferences

This may help explain why properties connected to transportation, mixed-use development, and strong community-centered downtown infrastructure continue attracting attention despite broader uncertainty in some sectors.

For investors, developers, business owners, and advisors alike, this environment can place increasing importance on understanding not just where the market is today, but how different sectors may evolve over time under changing conditions.

Investors and developers will be keeping a close eye on these market dynamics to see whether they continue reinforcing the current direction of development and investment activity across the region or begin shifting priorities and investment behavior in new directions.

Recent Posts
Archives