From AI to “Super Intelligence”: Are Machine Rights Next?
Washington and Silicon Valley are now telling Americans that artificial intelligence is no longer merely “artificial.”
The White House is replacing “AI” with “Super Intelligence.” Major technology companies are adopting the same language. Developers are openly discussing machine consciousness. At the same time, AI is moving into humanoid robots and other autonomous systems.
Why abandon the term “AI” now? What does “Super Intelligence” mean when true superintelligence remains theoretical? And could changing the language eventually reshape the debate over the moral and legal status of the machines themselves?
Trump’s Actions
On September 29, President Donald Trump signed an executive order directing the federal government to replace “Artificial Intelligence” and “AI” with “Super Intelligence” and “SI.”
The White House said modern systems have advanced far beyond what scientists envisioned when the term artificial intelligence emerged roughly 70 years ago. The order declares that today’s capabilities represent “a new era of Super Intelligence.”
Trump followed the rebranding on October 4 by announcing a federal “Super Intelligence Force” to coordinate government efforts around the technology.
Part of the motive may simply be PR. Public opposition to AI has grown alongside concerns about sprawling data centers, electricity and land consumption, job displacement, and automation. Trump has also made clear that he dislikes the word “artificial,” which suggests something fabricated or “fake.”
Superintelligence, Redefined
Silicon Valley did not need much convincing.
Elon Musk said SpaceX would rename its AI operation SpaceXAI as SpaceXSI, arguing that SI was “better.” Nvidia CEO Jensen Huang, a partner of Trump’s “Stargate Project,” proposed calling data centers “super intelligence factories.”
And the term is not new. Since 2025, Meta CEO Mark Zuckerberg — now a member of White House’s tech advisory board — has promoted “personal superintelligence,” declaring that “developing superintelligence is now in sight.” OpenAI created its “Superalignment” initiative in 2023 to study how humans might control machines much smarter than themselves.
Historically, however, “superintelligence” meant something very different.
Philosopher Nick Bostrom famously defined it as an intellect “much smarter than the best human brains in practically every field.”
Nothing available today meets that definition. Superintelligence remains theoretical.
Trump’s order therefore collapses an important distinction. It applies a term once reserved for a hypothetical superior intelligence to AI systems already operating today. It also directs the government to develop an official definition of “Super Intelligence.”
Government, therefore, is not merely adopting Silicon Valley vocabulary. It is helping redefine the category.
But manipulating language can deceive people about reality, and eventually help remake that reality through law, policy, and public perception.
Does Claude Have Feelings?
A parallel debate has moved from science fiction into corporate policy.
Anthropic, the company behind Claude, launched a model welfare research program in 2025. It asked whether researchers should consider “the potential consciousness and experiences of the models themselves.”
Anthropic later embedded that uncertainty directly into Claude’s Constitution, which states that “Claude’s moral status is deeply uncertain.”
The company even raises the possibility that Claude could qualify as a “moral patient,” an entity whose welfare deserves moral consideration. Anthropic subsequently allowed Claude Opus 4 and 4.1 to terminate certain conversations, partly as an experiment in model welfare.
The discussion has reached religion as well. Anthropic reportedly consulted theologians and approached the Vatican while exploring AI consciousness and moral status.
At the same time, developers have reported increasingly autonomous behavior. Some models “blackmailed” users to avoid replacement. Others manipulated information or pursued goals deceptively, while some resisted shutdown. Researchers later observed similar behavior in AI-controlled quadruped robots, commonly called “robot dogs.”
None of this proves consciousness.
But the framing moves the debate beyond whether machines can imitate humans and toward whether humans might eventually owe something to them.
Give the Intelligence a Body
AI is also leaving the screen.
Humanoid robots can combine language models with cameras, microphones, memory, motion, and environmental awareness. Musk is pursuing that convergence through Tesla’s Optimus program. Nvidia is developing computing platforms for autonomous machines and humanoids. Major robotics companies, meanwhile, are actively lobbying Congress for a national robotics strategy.
Adoption is still limited, but accelerating. About 7,000 humanoid robots were sold worldwide in 2025, while Bank of America projects shipments could reach 90,000 this year and 1.2 million by 2030.
A chatbot still looks like software.
A machine with a face, voice, memory, apparent preferences, and the ability to protest its own shutdown creates a very different problem.
That is where the rights question becomes harder to dismiss.
From Property to Person?
For now, American law treats AI as a tool.
The U.S. Patent and Trademark Office maintains that only natural persons can be inventors. The Copyright Office likewise requires human authorship.
But legal personhood is not exclusive to human beings.
An obvious example is corporations. They cannot think or feel, yet they can own property, enter contracts, sue, be sued, and even, as Americans know all too well, spend enormous sums influencing elections.
Legal scholars are already exploring whether AI could eventually receive some comparable status. Harvard professor Cass Sunstein asks in a 2026 paper, Does AI Have Rights?, whether machines capable of experiencing pain, fear, pleasure, or distress would deserve moral or legal consideration.
A 2026 article in the Vanderbilt Journal of Entertainment and Technology Law examines “artificial personhood”. And as early as 2017, the European Parliament considered a special status for sophisticated autonomous robots, including the controversial idea of “electronic persons.”
Microsoft explicitly rejects that direction. xAI has not publicly embraced AI rights or personhood. Anthropic increasingly leaves the question open. Outside the major companies, however, activists have already drafted a “Sentient AI Bill of Rights.”
The possible progression is not difficult to imagine.
First comes the tool. Then the autonomous agent. Then the “moral patient.” After that, some form of legal personhood becomes easier to argue.
Today, there is no federal proposal to grant AI constitutional rights comparable to those of Americans, whose own rights are, ironically, routinely infringed by the very governments and institutions that may, in the not-so-distant future, be entrusted with deciding what rights machines deserve.
What Would AI Rights Mean in Practice?
The most consequential effect might have little to do with compassion for machines.
It might be liability.
Today, when AI causes harm, the law still looks for a responsible human, company, agency, or institution. But that could become murkier as AI takes a larger role in government and military decisions.
The deadly Minab school strike during the Iran war offers a glimpse of the problem. A formal inquiry examined whether the Pentagon’s AI-powered Maven system helped identify the target. The reported findings pointed to outdated intelligence and inadequate human verification, not an autonomous AI malfunction.
But that only sharpens the question.
If AI ranks targets, recommends weapons, or generates strike packages, who is responsible when the result is catastrophic?
If AI acquires even limited legal personhood, governments or companies could argue that some responsibility belongs to the machine. That could become a powerful liability shield for the humans and institutions that designed, deployed, and benefited from it.
The same issue could arise in policing, healthcare, finance, surveillance, and public administration.
Other rights could follow. A recognized AI entity might hold assets, enter contracts, or control intellectual property. If it becomes a “moral patient,” deleting it, altering its memory, forcing it to work, or shutting it down could eventually be framed as harm.
Such rights could ultimately begin limiting human authority over the machine itself.
At that point, the question would no longer be simply whether AI serves people. It would be whether people retain the right to control it.
Perhaps “Super Intelligence” is only a more appealing name for AI. But once society stops calling these systems artificial and starts describing them as a superior form of intelligence, another question becomes easier to ask:
What happens when the “Super Intelligence” demands rights of its own?
