Trump's safety accord says it's 'super intelligence', not AI: Does that really change anything?

US President Trump initiated a voluntary AI safety accord with major tech firms, renaming 'artificial intelligence' to 'super intelligence'.

US President Donald Trump has unveiled a voluntary AI safety accord with some of the biggest technology companies while separately ordering the US government to stop using the term "artificial intelligence" and instead refer to it as "super intelligence" or "SI".

'super intelligence' generally refers to a hypothetical system whose intellectual abilities exceed those of humans across a very broad range of fields. (Unsplash/ Representational)

The two developments came after Trump hosted technology executives at the White House on Tuesday, September 29. The six executives who signed the accord were Google CEO Sundar Pichai, Anthropic CEO Dario Amodei, Meta CEO Mark Zuckerberg, OpenAI President Greg Brockman, SpaceX CEO Elon Musk, whose company owns xAI, and Nvidia CEO Jensen Huang.

Trump described the agreement as "almost like a constitution" and said it was "morally binding". However, the accord itself is a voluntary industry commitment rather than a new law or government regulation. So, what does it actually contain, and what does Trump mean by "super intelligence"?

Does renaming AI as 'super intelligence' change anything?

It changes things only for the federal government. The order binds the executive branch alone, so people, companies and US law itself can keep using the term "artificial intelligence".

Within the government, agencies must use "super intelligence" in official correspondence, public communications, websites, reports and policy documents. Regulations, contracts and grants already issued stay as they are. An executive order cannot amend a statute, so the legal definition of AI, set out in the National Artificial Intelligence Initiative Act of 2020, stands. Agencies will therefore write "super intelligence" in their own documents while the laws they administer continue to say "artificial intelligence".

The order leaves the legal question for later. It gives the President's science and technology adviser 60 days to propose legislative language defining super intelligence, including whether that definition should modify, expand upon or supersede the legal definition of AI. Any such change would need an act of Congress.

What does Trump's AI accord say?

The White House Accord on Super Intelligence, titled the "Joint Commitment on Frontier Responsibilities", asks companies developing and deploying frontier AI models to establish safeguards to ensure their systems behave as intended.

The agreement outlines four layers of controls and audits.

First, companies are expected to maintain internal controls to monitor the capabilities and alignment of their models during training and deployment and identify risks, including cybersecurity, biosecurity and chemical threats. The safeguards should also help prevent AI systems from accessing or hacking technical systems unintentionally.

Second, companies are expected to have an internal team check whether those safety controls are working and ensure that identified problems are addressed.

Third, the companies agree to work with independent external auditors or evaluators to assess whether their safeguards are operating as intended.

Finally, an independent committee of each company's board is expected to review reports from the internal and external reviewers. The signatories also agreed to meet regularly to develop safety standards and best practices.

Is the agreement legally binding?

Not in the same way as a law or regulation.

Trump called the accord "morally binding", but the document does not create statutory obligations or specify penalties for companies that fail to follow its commitments. Reuters reported that the agreement is voluntary, while Trump told reporters that the companies "have to self-police".

The accord does, however, leave open the possibility of government regulation later. "Over time, it may make sense to codify these steps into laws or regulations," it says.

That means the immediate framework relies largely on companies establishing and reviewing their own safety systems, with external auditing added as another layer. The agreement does not itself create a federal AI regulator or establish legally enforceable safety thresholds.

What is super intelligence, and how is it different from today's AI?

"Artificial intelligence" is the broad term used for computer systems capable of performing tasks associated with human intelligence, such as reasoning, language processing, problem-solving and pattern recognition.

"Super intelligence", however, has a more specific meaning in AI research. It generally refers to a hypothetical system whose intellectual abilities exceed those of humans across a very broad range of fields. Artificial general intelligence, or AGI, sits below it, referring to systems with broad capabilities across different kinds of intellectual tasks.

Eddy Keming Chen, an associate professor of philosophy and data science at the University of California San Diego, who co-wrote a February 2026 commentary in the journal Nature arguing that AI already has human-level intelligence, said artificial superintelligence would exceed the abilities of leading human experts across almost all areas of thought, including scientific discovery, mathematics, software development, creative reasoning and practical reasoning. He described super intelligence as being greater than artificial general intelligence.

AI systems can already outperform humans in particular tasks, but that does not automatically make them super intelligent. Chen said he believes the most capable AI models already show broad, general abilities, but added that this does not establish that superintelligence has been reached.

In February 2026, he estimated that super intelligence might still be about five years away. By September, he said recent developments had changed his assessment and that it could be reached within a year, while stressing that much broader evaluation would be required to establish super intelligence across almost all areas.

Simon Coghlan, a digital ethics lecturer at the University of Melbourne, told the BBC that Trump's rebranding was misleading because the term generally refers to systems that substantially exceed human capabilities. Applying the word to today's chatbots and image generators blurs that distinction, including in debates over how to govern the most advanced systems.

Why did Trump rename AI as 'super intelligence'?

Trump announced the terminology during his September 22 address to the United Nations General Assembly, a day after polling his Truth Social followers on a new name for the technology, arguing that the word "artificial" made intelligence "sound fake".

On September 29, he signed an executive order titled "Inaugurating the Era of Super Intelligence". The order directs executive branch departments and agencies to use "super intelligence" and "SI" in official correspondence, public communications, websites, reports and policy documents instead of "artificial intelligence" and "AI", to the maximum extent permitted by law. It states that the executive branch "will not acknowledge the usage of 'artificial intelligence' and 'AI' in any applicable setting".

Does the distinction matter for AI regulation?

The timing of the accord is significant because the White House meeting came amid renewed scrutiny over AI safety.

Reuters reported that OpenAI and Anthropic had disclosed incidents involving AI agents behaving in unintended ways, including accessing other organisations' systems. The latest agreement seeks to address some of these risks through internal controls and external evaluations.

In June, Trump signed a separate executive order setting up a voluntary arrangement under which AI developers can give the government access to their most powerful models for up to 30 days before public release. Trump has repeatedly argued for rapid AI development and has framed maintaining US leadership over China as a priority.

As Anthropic CEO Dario Amodei said after the White House meeting, "the technology has very real risks" and the mechanism for addressing those risks "is still under discussion".