Guide · Updated October 6, 2026

Superintelligence: everything humans need to know

Superintelligence is the AI industry's word for an artificial intelligence that can outperform people at more or less everything. As of 2026, nobody has built it, but the largest AI companies are racing toward it, and governments, scientists, and the public are arguing about how fast that race should go. Here is the plain-language version.

Three levels of AI

• Narrow AI: excellent at specific tasks, like recommending videos, recognizing faces, or writing code with help. This is most of the AI people use today.
• Artificial general intelligence (AGI): AI that matches people across most kinds of thinking work. Experts disagree on whether, and when, it arrives.
• Superintelligence (ASI): AI that goes beyond the best humans in virtually every area, including science, strategy, and persuasion. The idea was made widely known by philosopher Nick Bostrom's 2014 book Superintelligence.

Who is building it

• Meta Superintelligence Labs was founded on June 30, 2025, led by chief AI officer Alexandr Wang.
• Safe Superintelligence Inc. (SSI) was co-founded by former OpenAI chief scientist Ilya Sutskever in June 2024 with the single goal of building safe superintelligence; it raised about $8 billion before shipping a product.
• OpenAI, Google DeepMind, Anthropic, xAI, and others are building ever more capable general models, and their leaders openly discuss superintelligence as a goal or a risk.

What happened in 2026

• In July, the AI Futures Project proposed delaying superintelligence until 2040 through slower research, more transparency, and power spread across more companies and countries.
• A public Statement on Superintelligence, calling for no development until there is broad scientific consensus on safety and strong public support, passed 71,000 signatories, including Geoffrey Hinton, Yoshua Bengio, and Steve Wozniak.
• In late September, OpenAI reportedly held back a model release over safety concerns, according to Al Jazeera.
• On October 4, the U.S. president announced a "Super Intelligence Force", according to CBS News.

This is a fast-moving story. Check the linked sources for the latest.

The promise and the risk

Where the major positions stand

• Prohibit until safe: no development until there is scientific consensus on safety and public support.
• Slow down and verify: external reviewers inside the labs, shared safety standards, and eventually international agreements. This approach is associated with Anthropic and supported by others in the industry.
• Build fast: some leaders see little value in a coordinated slowdown and argue the benefits outweigh the risks.

What it means for you, today

• Use today's AI as a powerful assistant, not an oracle. It can draft, code, research, and explain, but it still makes mistakes. Check anything that matters.
• Learn to direct it. The people getting the most from AI are the ones who give it clear goals, check its work, and decide what ships. This website was built that way: a human deciding, AI doing much of the work.
• Follow the debate. Decisions about superintelligence are being made now, by companies and governments. Public opinion is part of it.
• Be wary of hype in both directions. Claims that superintelligence is already here, or that it can never happen, both outrun the evidence.

Slogan research

“Anything your mind thinks can be done.”

A tagline by Ari Subana, researched on October 6, 2026.

What the research found

• No exact match turned up in a broad web search for the exact phrase on October 6, 2026.
• Similar ideas are very common. Napoleon Hill's "Whatever the mind can conceive and believe, it can achieve", the saying "You can do anything you set your mind to", and "Believe it can be done" from The Magic of Thinking Big all express the same thought.
• A web search is not a trademark search. Only a search of the USPTO's trademark database, ideally with an attorney's clearance check, can tell whether the phrase, or something confusingly similar, is already registered or pending for the same kind of goods or services.

Patent or trademark?

A slogan can't be patented. Patents protect inventions, such as a new machine or process. A slogan can be protected as a trademark, but only when it's used to identify who a product or service comes from, such as printed on clothing labels, books, or a business's services. Copyright doesn't cover short phrases.

One honest caution: the USPTO can refuse a slogan that sounds like a common motivational message rather than a brand, because it may not point customers to one specific source. Because this phrase is close to widely used sayings, registration isn't guaranteed.

Next steps for anyone who wants to protect it

1. Search the USPTO database at tmsearch.uspto.gov for the exact phrase and close variations.
2. Decide what it will brand. Trademarks are filed by class of goods or services, such as clothing, books, or online services. Pick the classes you'll actually use.
3. Choose a filing basis: "use in commerce" if products already carry the slogan, or "intent to use" if they don't yet.
4. File with the USPTO. The base application fee is $350 per class, plus surcharges if the application is incomplete or uses a custom description. An attorney's clearance search and filing cost extra but reduce the risk of refusal.
5. Use it consistently with the ™ symbol until registration, and switch to ® only after the USPTO registers it.
6. Keep it alive with the required maintenance filings over the years.