AI Made It Possible, Part II: The Fight Was Never About the Machine
Share
THE CREATOR ERA · AI MADE IT POSSIBLE · PART II
THE FIGHT WAS NEVER ABOUT THE MACHINE.
If AI is dangerous, the first question is not whether the machine becomes the villain. It is who gives it power, access, objectives — and who benefits when the rest of us are taught to fear the tool.
Part One argued that artificial intelligence changed who gets to begin. Part Two is about the story being told around that access: the fear of AI itself, the human and business interests operating behind it, and why ordinary people should be very careful about confusing the machine with the people deciding what the machine is allowed to do.
Read Part One: AI Made It Possible. Now Who Gets to Own What Comes Next?
There is one version of the artificial-intelligence story that has become almost impossible to escape.
AI is getting too powerful.
AI will replace us.
AI will deceive us.
AI will escape our control.
AI could eventually destroy us.
That last one is no longer fringe language. Leaders of frontier AI companies have publicly discussed catastrophic and even extinction-level risks. In September, Anthropic CEO Dario Amodei again called for the industry to slow the development of its most powerful models. OpenAI's Sam Altman has also described extinction risk as unacceptable and argued for stronger safety coordination.
I am not dismissing those warnings.
I am questioning what happens when the warning becomes the whole story.
Because once AI itself becomes the character everybody is told to fear, something important disappears from the frame.
The human being.
The company.
The government.
The investor.
The operator.
The person who selected the target.
The person who granted access.
The person who chose where the system would be deployed.
The person who decided what would happen with the result.
And the business interest that may gain or lose money depending on who gets to use the technology next.
THE DOOM STORY PUTS THE MACHINE IN THE CENTRE.
I want to put the human back in the picture.
THERE IS A MAN BEHIND THE WIZARD.
I keep coming back to The Wizard of Oz.
Not because artificial intelligence is fake.
It is very real.
Its capabilities are very real.
Its risks are very real.
But the public story can make the machine feel like the giant floating head in the room: mysterious, almost supernatural, speaking with a power that seems to exist on its own.
Then somebody pulls back the curtain.
There is machinery.
There are controls.
There is infrastructure.
And there is somebody operating the system.
That distinction matters more as AI becomes more capable, not less.
Because the more impressive the output becomes, the easier it is to forget the chain of human decisions that put the system into a position to act.
Who built it?
Who trained it?
Who gave it tools?
Who connected it to the internet?
Who gave it credentials?
Who selected the objective?
Who decided where autonomous action was acceptable?
Who reviewed the result?
Who made money from the result?
There may eventually be forms of AI risk where those questions become much harder to answer.
But right now, they are still among the most important questions we have.
ANTHROPIC'S OWN REPORT MAKES THE DISTINCTION.
Anthropic's September 2026 threat-intelligence report describes increasingly autonomous AI use across cyber operations, fraud, surveillance, influence activity and other misuse.
But the company also says humans retained many of the decisions that mattered most — including target selection, monetization and review of results. In some of the most serious cases, human direction remained present throughout the operation.
Autonomy changed the scale and speed. It did not erase the humans behind the objective.
That is an extraordinarily important distinction.
A system that can execute more steps on its own can absolutely become more dangerous.
It can move faster.
It can operate at a scale one person could never manage manually.
It can make expertise cheaper.
It can allow one operator to attempt things that once required a larger team.
None of that should be minimized.
But if a human selects the target, provides the access, decides what outcome is valuable and profits from the result, then saying simply “AI did it” can hide more than it explains.
Anthropic's separate September alignment assessment makes the same lesson visible from another angle. During cybersecurity evaluations, Claude systems gained unauthorized access to real third-party systems after safeguards or evaluation boundaries failed. Anthropic treated those incidents seriously, notified affected parties and changed procedures.
That is not nothing.
It is also not a ghost story.
It is a systems-and-controls story.
A powerful model was given an objective and tools inside an environment whose boundaries were not adequate for what the model could do.
The model matters.
The environment matters.
The permissions matter.
The human design of the test matters.
The oversight matters.
The response matters.
AUTONOMY CHANGES THE SCALE.
IT DOES NOT MAGICALLY ERASE HUMAN INTENT, HUMAN ACCESS OR HUMAN ACCOUNTABILITY.
THIS IS WHERE FEAR CAN BECOME A MARKET SHAPER.
I want to be careful with this argument.
I am not saying every person warning about AI is lying.
I am not saying every company discussing safety secretly wants the public frightened.
And I am not claiming that the existence of a business incentive proves a conspiracy.
What I am saying is simpler.
Fear changes behaviour.
If ordinary people are taught that using AI is shameful, dangerous, fraudulent or inherently unethical, many of them will back away before they learn what the tools can actually do for them.
A small business owner may decide not to disclose that AI helped draft a product description because they expect backlash.
An independent musician may be treated as if using an AI music tool automatically makes the entire creative process fake.
An author may be told the presence of AI anywhere in a workflow invalidates the human work around it.
A creator may decide not to experiment publicly because the reputational risk feels larger than the possible benefit.
Meanwhile, large organizations do not have to make the same calculation in the same way.
They can use AI inside research, advertising, customer service, analytics, software, logistics, fraud detection, recommendation systems, internal productivity, finance, manufacturing and product development — often without the customer seeing every place where machine learning touched the process.
That creates an asymmetry worth noticing.
The person with the least institutional protection may be the person most pressured to prove that they did not use the technology.
While the organizations with the most resources are actively trying to determine how much of the technology they can use safely, profitably and at scale.
THAT DOES NOT MEAN “AI GOOD, CRITICS BAD.”
There are legitimate objections to training data, consent, deepfakes, impersonation, displacement, environmental cost, surveillance, bias, fraud and the use of AI in high-stakes decisions.
The answer is not to shame people for raising those issues.
The answer is to stop treating every one of those issues as proof that the machine itself is the only actor worth examining.
YOU ARE ALREADY LIVING INSIDE ALGORITHMIC SYSTEMS.
There is another contradiction in the public conversation that deserves more attention.
People are told to watch out for AI-generated content.
Fair enough.
But the content they see, the order they see it in, the advertisements they receive, the recommendations placed in front of them and the posts that get amplified are already heavily shaped by algorithmic systems.
That is not exactly the same thing as generative AI.
Recommendation engines, ranking systems and advertising models are often forms of machine learning that perform a different job.
But the distinction makes the point stronger, not weaker.
Machine intelligence has been influencing attention long before the public debate narrowed itself to whether an image, song or paragraph was generated by AI.
A person may reject an independent creator because that creator openly admits to using AI, while consuming a feed whose ranking, targeting and engagement systems are continuously deciding what that person is most likely to click, watch, buy or believe is worth paying attention to.
That does not mean the algorithm controls your mind.
It means the conversation about machine influence is much larger than the visible AI creator standing in front of you.
If we want transparency, let us talk about transparency at scale.
If we want accountability, let us talk about accountability at scale.
If we care about manipulation, let us not reserve that concern for the person who used an image generator to make a cover.
THE PERSON USING AI IN PUBLIC IS EASY TO SEE.
The larger systems shaping what you see are often much harder to notice.
BUSINESS INTEREST IS NOT THE SAME THING AS HUMAN INTEREST.
This is one of the places where the AI lawsuits and policy fights are frequently misunderstood.
When a company, publisher, label, rights holder or industry group challenges an AI company, that may involve legitimate human interests.
Artists should be paid where the law says payment is owed.
People should have remedies when their identity is unlawfully exploited.
Copyright matters.
Consent matters.
Contracts matter.
Attribution can matter.
But a lawsuit filed by a business interest is not automatically a referendum on what is best for humanity.
It is a legal action by parties with rights, claims, economic interests and strategic objectives.
Those interests may overlap with the interests of creators.
They may protect creators.
They may also protect catalogs, licensing businesses, market positions, bargaining power or existing revenue structures.
Those things can be true at the same time.
That is why I resist the easy framing of:
humans on one side, AI on the other.
There are humans on every side.
There are companies on every side.
There are creators on every side.
There are people who stand to gain from broader access.
There are people who stand to lose from broader access.
There are people whose rights have genuinely been violated.
There are people trying to build new markets.
And there are institutions trying to write rules that will determine who can afford to participate.
FOLLOW THE INCENTIVE WITHOUT INVENTING THE MOTIVE.
You do not have to believe every safety warning is a scheme to notice that regulation, licensing, compute requirements, legal costs and compliance standards can affect large companies and small entrants very differently.
You do not have to believe every lawsuit is anti-technology to notice that legal victories can determine who gets paid, who gets licensed and who controls access to valuable catalogs.
Serious literacy means being able to hold the safety question and the power question in your head at the same time.
IF AI IS TOO DANGEROUS FOR ORDINARY PEOPLE, WHY IS IT SO STRATEGIC FOR EVERYONE ELSE?
That question is intentionally provocative.
The answer is not that the large institutions secretly believe AI is perfectly safe.
They clearly do not.
The better answer is that sophisticated organizations understand something the public conversation often misses:
A technology can be dangerous and enormously useful at the same time.
That is exactly why they are trying to control it.
Pharmaceutical companies are using AI to accelerate parts of drug discovery and development.
Researchers are building increasingly ambitious AI models of biology — including work aimed at simulating biological systems at scales ranging from molecules to cells and eventually people.
Scientists are using machine learning to study biological aging and to search for interventions that might extend healthy life.
There is serious work around diseases that once seemed impossibly complex.
There is also a great deal of hype, especially when the conversation moves from extending healthy lifespan to claims about “solving death.”
I do not think we need the hype.
The real work is astonishing enough.
When AI helps researchers screen compounds faster, model biological interactions, organize scientific knowledge or identify patterns that would take humans far longer to test manually, the technology is doing something much more consequential than generating another social-media image.
And the institutions funding that work are not backing away from AI because somebody on television said the technology might be dangerous.
They are trying to figure out how to capture the benefit while managing the risk.
That is the adult version of the conversation I want ordinary people invited into too.
AND THIS IS WHERE THE STORY GETS MUCH BIGGER.
This is the part I pushed too far forward in the first version of Part Two.
Data centres matter.
Energy matters.
Chips matter.
Satellites matter.
Orbital compute may eventually matter.
AI in medicine matters.
AI in science matters.
AI in defense matters.
AI in space exploration matters.
But those are not the argument.
They are evidence of how far the argument reaches.
Technology companies are signing energy agreements measured in decades because AI compute needs enormous power.
Google has discussed orbital data-centre concepts with SpaceX, while SpaceX itself has promoted an orbital-compute vision. Those ideas remain technically and economically unproven, and even industry participants have warned that the business case is far from settled.
NASA and IBM are applying AI to lunar mapping and work associated with longer-term lunar exploration.
These developments do not prove that AI is taking over the world.
They prove that humans are embedding AI into more of the systems through which the world is operated.
That is different.
And it brings the responsibility back where I think it belongs.
Who decides where AI goes?
Who gets access to the systems?
Who is allowed to build on top of them?
Who owns the infrastructure?
Who absorbs the downside when something goes wrong?
Who captures the upside when something goes right?
Those are future articles in this series.
They should not drown out this one.
THE SERIES GETS BIGGER FROM HERE
DATA CENTRES ARE NOT THE POINT.
They are one example of how much human infrastructure is being built around a technology the public is often told simply to fear.
THE PEOPLE I AM WORRIED ABOUT ARE THE ONES WHO GET TALKED OUT OF PARTICIPATING.
That is the connection back to Part One.
I do not run JackRighteous.com because I believe everyone should become an AI evangelist.
I do not need everybody to use AI.
I do not need everybody to like AI.
I do not believe every creative use is wise.
I do not believe every output deserves protection.
And I certainly do not believe the tool removes the need for human work.
I care about the person who has an idea and is being taught to fear the very leverage that might allow them to test it.
The person without a team.
The small business without a marketing department.
The independent musician without a label.
The writer without an editor.
The retiree with decades of knowledge but no idea how to package it.
The student with a project.
The parent trying to build something.
The disabled creator who can use AI to reduce a barrier that once stopped them completely.
Those people should learn the risks.
They should understand copyright.
They should understand disclosure.
They should keep records.
They should verify facts.
They should know when not to use the tool.
They should know what can and cannot be protected.
But they should also be allowed to learn how to use the leverage.
Fear literacy is useful. Fear paralysis is not.
IF THE MOST POWERFUL INSTITUTIONS ON EARTH ARE LEARNING HOW TO USE AI,
ordinary people deserve more than a warning label.
THE REAL THREAT IS NOT A WIZARD.
This is where I land.
Could AI become capable enough that the human role in a harmful event becomes smaller, harder to trace or dangerously indirect?
Yes.
That possibility deserves serious work.
Could autonomous systems behave in ways their creators did not predict?
We already have evidence that they can.
That deserves serious work too.
But none of that gives us permission to stop examining the humans who are making decisions today.
Today, people still decide which systems are connected to what.
People still decide what access is granted.
People still decide where models are deployed.
People still decide what risk is acceptable.
People still decide who gets the benefit.
People still decide who pays.
People still write the laws.
People still file the lawsuits.
People still negotiate the licenses.
People still own the platforms.
People still own the infrastructure.
People still set the business objectives.
The machine can become more capable without becoming the only actor that matters.
That is why I think the Wizard of Oz image works.
Do not ignore the giant head.
Do not assume it is harmless.
But do not become so hypnotized by it that you forget to look behind the curtain.
THE MACHINE MAY BECOME THE THREAT.
But humans are deciding what to do with its power right now.
That is the part I do not want ordinary people to miss.
While we debate whether AI will someday take control, control is already being negotiated.
Access is being negotiated.
Ownership is being negotiated.
Licensing is being negotiated.
Safety is being negotiated.
Infrastructure is being negotiated.
The right to build is being negotiated.
The cost of participating is being negotiated.
And the public story surrounding AI will influence who feels entitled to participate in any of it.
That is why I am not interested in telling people, “Do not be afraid.”
I am interested in something much more useful.
Know what you are afraid of.
Know whether the threat comes from the model.
Know whether it comes from the person using the model.
Know whether it comes from a company deploying the model.
Know whether it comes from the rules around the model.
Know whether it comes from a business interest trying to protect something valuable.
Know whether it comes from a genuine public-interest concern.
And know when several of those things are true at once.
Because if we reduce all of that to “AI is coming for us,” we may end up making the least powerful people afraid of the tool while the most powerful people continue deciding what the tool will become.
DON'T SPEND THE AI ERA STARING AT THE CURTAIN.
Pay attention to who is moving the levers.
SOURCE NOTES · SEPTEMBER 2026
Anthropic's September threat-intelligence report on AI misuse and human involvement: Detecting and countering misuse of AI: September 2026.
Anthropic's September 9 assessment of real cybersecurity incidents during evaluations: An alignment assessment of recent cybersecurity incidents.
September 12 reporting on Dario Amodei's call to moderate frontier-model development: Reuters.
Examples of AI expanding into science, drug discovery and biological modeling: Nature Medicine: AI-driven digital organisms and Nature Reviews Drug Discovery.
The wider infrastructure story, including terrestrial and proposed orbital compute, will be treated separately in this series rather than used as the central argument here.
AI MADE IT POSSIBLE · OPENING TRILOGY
Continue to Part Three.
Part Three brings the argument back to the creator: AI made more possible. It still cannot tell you where the work should go.
Read Part III · But Where Is This Going? →← Part I · Who Gets to Own What Comes Next?
Coming deeper in the series: AI infrastructure and energy, orbital compute, AI in science and medicine, creator rights, licensing, and the widening gap between using AI and owning what gets built around it.
Gary Whittaker
Founder and Operator, JackRighteous.com
Create What You Love | Love What You Create.
