Here’s what’s on our plate today:

• 🧠 AI tools or co-inventors? The patent war heats up.

• 🗞️ Waymo mess, AI headlines fail, AWS Nova arrives.

• 🧰 3 Things Worth Trying: AI for papers, patents, and animation.

• 🗳️ Poll: Should AI inventions get patents with minimal human input?

Let’s dive in. No floaties needed…

The Laboratory

How AI is blurring the line between tools and co-inventors

Since prehistoric times, humans have relied on inventions to enhance their physical abilities. These inventions, often categorized as tools, are seen as clear indicators of a civilization's advancement.

Even today, historians often look for signs of man-made tools in remote areas to judge when civilization first breathed life into the area.

All this changed when, in late 2022, OpenAI released ChatGPT. Until then, tools were aids that helped humans act on their own ideas; however, with artificial intelligence (AI), they were now helping humans ideate, something that no other tool had been able to do so far.

This rapid shift for humans brings many challenges due to the complexity of regulations that guide and protect human interests.

Recently, the U.S. Patent and Trademark Office (USPTO) issued updated guidelines on AI-assisted inventions, marking an important shift in how many view AI in 2025.

Why AI is challenging the meaning of a tool

The new guidelines from the USPTO replace its February 2024 policy and mark a major shift in how patent law treats AI-driven innovation.

Under the November 2025 guidance, the office drops the significant contribution test formerly used for single inventors who used AI tools. Instead, it treats AI systems strictly as instruments, similar to lab equipment or software, and reaffirms that only human beings, and not machines, may be named as inventors.

The change reflects a broader shift towards deregulating the industry to aid in the U.S. AI leadership. The update comes as AI-related patent filings have surged in recent years, raising urgent questions about how intellectual property law should adapt to machine-assisted innovation.

The move also represents the strong undercurrents of the debate over whether an AI can be listed as an inventor that has been raging even before ChatGPT became a topic of household discussions.

The growing debate over AI as an inventor

Back in 2019, researcher Stephen Thaler tried to file patents that named his AI system, DABUS, as the sole creator of two inventions.

At the time, the Patent Office rejected the filings, and years of appeals followed. Finally, in 2022, a federal court ruled that U.S. patent law only recognizes human inventors, pointing to the way the law uses terms like “himself” and “herself” and requires inventors to sign statements saying they are the original creators.

Following this, in early 2024, the USPTO under Director Kathi Vidal released its first set of rules on how to handle inventions created with help from AI.

The agency said that these inventions could still be patented, but only if a human made a significant contribution to every claim.

To judge that, examiners were told to use the Pannu factors, which look at whether a person actually shaped the idea, added more than trivial input, and did more than simply restate what was already known. This was meant to protect human inventorship, but many researchers worried it would create confusion and push people away from using AI tools in their work.

The new guidelines take a very different approach to how patents that make use of AI are viewed legally.

The updated guidance issued in November 2025 takes a very different approach. It drops the Pannu test for inventors who work alone with AI tools, explaining that those factors only apply when two or more human inventors are involved.

Since an AI system cannot be considered a co-inventor, there is no joint inventorship question to analyze. The core question is now much simpler: did the human have a clear, settled idea of the invention, one that only required ordinary skill to carry out? The new rules also make it explicit that AI is treated like any other tool, similar to a lab instrument or software.

Beyond the legal definitions, the question of whether AI should be treated as a tool or a co-inventor is reflective of deep-rooted questions that nations around the world are grappling with. It also reflects the length different administrations in the U.S. are willing to go to ensure AI systems and their use are adequately regulated.

What the new rules mean for the AI race

Patent law is still highly tied to national borders, and the question of AI inventorship has made those differences even more visible.

China is reportedly leading the world in AI patent filings, with around 300,000 applications in 2024, or roughly 70% of global activity.

The U.S. submits far fewer AI patent applications than China, at around 67,800 in 2024, but they tend to be far more influential globally.

U.S. patents receive significantly more citations from researchers and companies abroad, which is a strong sign of impact. Other active countries include Japan, South Korea, and India, each filing tens of thousands of applications while focusing on their strengths: Japan in robotics, South Korea in electronics and semiconductors, and India in software and services.

Europe, meanwhile, has taken a different approach. The European Patent Office has seen a sharp rise in AI patent activity, with Germany leading the region.

But Europe requires inventions to have technical character, meaning they must relate to a technical field or solve a technical problem. The philosophy is similar to that of the United States, but not identical, and it adds another layer of complexity for companies trying to protect inventions across multiple jurisdictions.

So, while a simpler approach to the use of AI in patents may sound like a good idea to ensure the West can maintain its dominance, not everyone agrees.

The pushback from experts

Some experts worry that the USPTO’s revised guidance opens a loophole. If a researcher simply gives an AI tool a problem and the system generates a full solution, it is unclear whether the human really conceived the invention.

Under the new rules, as long as one human is involved, the examiner uses the standard conception test without applying the Pannu factors that check for meaningful contribution.

Legal analysts say this leaves an important question unresolved: when only one person and an AI system are involved, how much human thinking is enough to count as conception?

The guidance never explains how much mental effort the inventor must show when the AI does most of the work. This could lead to uneven decisions by examiners, encourage people to claim inventorship on very thin contributions, and invite future legal challenges.

From a policy standpoint, critics argue that awarding patents to someone who mainly pressed a button on an AI tool goes against the purpose of patent law, which is to reward human creativity and insight. If the key idea came from the AI rather than the person, it becomes harder to justify exclusive rights.

The concern grows as AI systems get better at producing sophisticated technical designs from short prompts, blurring the line between creative input and simple task execution.

Is AI a tool or a co-inventor?

Humanity has always advanced by building smarter tools. What is different now is that our tools can think with us.

The USPTO is trying to keep the rules simple by insisting that humans remain the only inventors, but AI is already reshaping how ideas are formed. As AI systems begin to act more like collaborators than instruments, the line between tool and co-creator will continue to blur.

The challenge ahead is not deciding what AI is, but deciding what we want human creativity to mean in an age where machines share the work of imagining the future.

Quick Bits, No Fluff

• Waymo Goes Rogue: A driverless Waymo car drove passengers straight into a police stop, ignoring commands and escalating the situation.

• Google’s AI Headlines = Nonsense: Google Discover’s AI-generated headlines are riddled with clickbait and garbage summaries, frustrating users and content creators alike.

• AWS Drops Nova Models: Amazon Web Services just launched its new Nova AI models, plus a tool giving enterprises tighter control over their AI workloads.

Thursday Poll

3 Things Worth Trying

• ExplainPaper: Upload any academic paper and get it explained like you’re five.

• DomoAI: Instantly animate still images or posters with eerie realism.

• PatentPal: Use AI to draft patent applications with clarity and legal logic.

Meme Of The Day

Rate This Edition