On Monday, the Biden administration unveiled sweeping restrictions on the export of artificial intelligence chips and related technologies as President Joe Biden prepares to leave office.

Announced just days before Donald Trump’s transition as the President, the “Interim Final Rule on Artificial Intelligence Diffusion” seeks to maintain U.S. dominance in AI while preventing adversaries such as from exploiting advanced systems for malicious purposes.

“To enhance U.S. national security and economic strength, it is essential that we do not offshore this critical technology and that the world’s AI runs on American rails,” according to a statment from the White House.

The restrictions impose caps on the number of advanced graphics processing units and other AI-related technologies that can be exported to most countries. 

While the U.S. and 18 of its closest allies, including the UK and Japan, are exempt, countries like China, Russia, Iran, and North Korea remain under strict bans. 

The regulations also introduce new licensing requirements for exports to over 120 nations, with provisions for foreign governments to sign agreements for eased restrictions.

Entities meeting rigorous security standards can gain special statuses, such as Universal Verified End User (UVEU) or National Verified End User (NVEU), allowing them to benefit from advanced GPUs and scale AI capabilities responsibly. 

UVEU entities can allocate up to 7% of their global AI capacity outside close allies, while NVEU entities can purchase GPUs equivalent to 320,000 advanced units over two years.

Smaller, low-risk chip orders—often used by universities and research institutions—will bypass the licensing process entirely.

Pushback

The announcement has drawn sharp criticism from the tech industry. 

Chipmaker giant Nvidia criticized the policy as “unprecedented and misguided,” warning that it could undermine U.S. innovation and global competitiveness.

“While cloaked in the guise of an ‘anti-China’ measure, these rules would do nothing to enhance U.S. security,” Nvidia’s president of government affairs, Ned Finkle, said in a statement.

Currently, Nvidia sells AI chips to China, but they are scaled-down versions designed to comply with U.S. export restrictions imposed in 2022. 

These chips with reduced computational power are produced to meet demand in the Chinese market without breaching U.S. security policies.

The new rule also caps chip exports to other non-allied nations at 50,000 GPUs per country, ensuring that U.S. technology supports legitimate uses, such as healthcare and education, without enabling adversaries. 

Nations that align their AI and export control policies with the U.S. can double their chip caps via government-to-government agreements under the new rules.

The rules are set to take effect mid-May, providing time for adjustments under Trump’s leadership following his inauguration on January 20.

“We hope that the next administration takes full advantage of those 120 days to listen to experts, industry, industry players, partner countries, consider their input, and I fully expect the next administration may make changes as a result of that input,”  US Secretary of Commerce Gina Raimondo said.

Edited by Sebastian Sinclair

Mark Zuckerberg approved using pirated books to train Meta AI, even after his own team warned the material was illegally obtained, a group of authors allege in a recent court filing.

The allegations come from a copyright infringement lawsuit filed by a group of authors including the comedian Sarah Silverman, Christopher Golden, and Richard Kadrey in a California federal court in July 2023. The group claimed Meta misused their books to train its Llama LLM, and they’re asking for damages and an injunction to stop Meta from using their works. The judge in the case dismissed most of the author’s claims in November of that same year, but these recent allegations may breathe new life into the legal dispute.

“Meta’s CEO, Mark Zuckerberg, approved Meta’s use of the LibGen dataset notwithstanding concerns within Meta’s AI executive team (and others at Meta) that LibGen is ‘a dataset we know to be pirated,'” lawyers for the plaintiffs said in a Wednesday filing. Despite these red flags, the lawsuit alleges that, “after escalation,” Zuckerberg gave the green light for Meta’s AI team to proceed with using the controversial dataset.

Representatives for Meta did not immediately respond to Decrypt’s request for comment.

LibGen, short for Library Genesis, is an online platform that provides free access to books, academic papers, articles, and other written publications without properly abiding by copyright laws. It operates as a “shadow library,” offering these materials without authorization from publishers or copyright holders. It currently hosts over 33 million books and over 85 million articles.

The lawsuit alleges Meta tried to keep this under wraps until the last possible moment. Just two hours before the fact discovery deadline on December 13, 2024, the company dumped what plaintiffs describe as “some of the most incriminating internal documents it has produced to date.”

Meta’s own engineers seemed uncomfortable with the plan, according to statements in court filings. The group of authors allege internal messages show Meta engineers hesitated to download the pirated material, with one noting that “torrenting from a [Meta-owned] corporate laptop doesn’t feel right (smile emoji).” Nevertheless, they proceeded to not only download the books but also systematically strip out copyright information to prepare them for AI training, the lawsuit claims.

The latest filings in the lawsuit paint a picture of a company fully aware of the risks: One internal memo warned that “media coverage suggesting we have used a dataset we know to be pirated, such as LibGen, may undermine our negotiating position with regulators.” Yet Meta went ahead anyway, both downloading and distributing (or “seeding”) the pirated content through torrenting networks by January 2024, according to the lawsuit.

When questioned about these activities in a deposition, Zuckerberg appeared to distance himself from the decision, testifying that such piracy would raise “lots of red flags” and “seems like a bad thing.”

The court documents also suggest that Meta’s approach to handling copyrighted information paid more attention to model training than copyright rules. According to the filing, one engineer “filtered […] copyright lines and other data out of LibGen to prepare a CMI-stripped version of it to train Llama.” This systematic removal of copyright information could strengthen the authors’ claims that Meta knowingly tried to hide its use of pirated materials.

The revelations come at a crucial time for Meta’s AI ambitions. The company has been pushing hard to compete with OpenAI and Google in the AI space, with Llama 3.2 being the most popular open source LLM, and Meta AI being a solid free competitor to ChatGPT with similar features.

Most of these AI companies are facing legal battles due to their questionable practices when it comes to training their large language models. Meta was already sued by another group of authors for copyright infringements, OpenAI is currently facing different lawsuits for training its LLMs on copyrighted material, and Anthropic is also facing different accusations from authors and songwriters.

But in general the tech entrepreneurs and creators have been up in arms ever since generative AI exploded in popularity. There are currently dozens of different lawsuits against AI companies for willingly using copyrighted material to train their models. But as with most things on the bleeding edge, we’ll have to wait and see what the courts have to say about it all.

As CES 2025 unfolds, one thing is clear—artificial intelligence is everywhere.

From TVs to vacuum cleaners, consumer electronics companies are racing to showcase the new AI features, sometimes shoehorned, into their products.

Some of these AI-powered products are impressive, while others stretch the meaning of “artificial intelligence” to its limits.

Here’s a look at some of the most eye-catching and occasionally absurd AI-powered products at CES so far.

The Roborock Saros 270: The robot vacuum claw machine

What it Does:

Developed by Roborock, the Saros 270 is a robot vacuum equipped with a robotic arm that moves small objects out of the way while cleaning. It has a charging station and can lift up to 300 grams, or 0.66 pounds.

Why It’s Absurd:

So, it’s cool, but the Saros 270 is limited by its size, making it useful for only picking up small toys and lightweight shoes and socks.

The claw it uses to pick up objects is only good for small and lightweight objects. Beyond that, what’s the point?

Unless the Saros 270 can carefully deposit those objects in a basket like a carnival claw machine, you’re left with slightly rearranged clutter. Fun? Yes. Practical? That’s debatable.

The SwitchBot K20+ Pro: The Swiss Army Knife of robot vacuums

What it does:

The SwitchBot K20+ Pro is another autonomous household robot. It isn’t just a vacuum—it’s an all-in-one home helper.

This robot can carry a humidifier, maneuver between rooms, and even collect floating pet hair from the air. Need to cool down at night? Attach a fan. Want a drink delivered? Add a shelf and let it roll your snacks around.

Why it Stands Out:

Its versatility is impressive. Unlike standard robot vacuums, the K20+ Pro feels more like a quirky butler on wheels.

It’s playful and genuinely useful—if you’re into the idea of your vacuum multitasking as a drink coaster.

Samsung Vision AI: AI for your TV

What it does:

Samsung’s Vision AI is part of their evolving “SmartThings” ecosystem.

Samsung’s Smart TVs now integrate AI to recognize their surroundings, adjust to user preferences, and offer generative AI features like creating digital art for wallpapers and screen savers and providing real-time subtitle translation during live broadcasts.

Why It’s Absurd:

While the tech sounds fancy, AI-generated wallpapers and live translation feel more like marketing gimmicks than necessities.

Plus, the more connected your TV is to other smart appliances, the bigger the cybersecurity risk. Do we really need another entry point for hackers in our living rooms, this time powered by AI?

Omnia Smart Mirror: Your reflection and health hub

What It Does:

Making the rounds at CES, the Omnia Smart Mirror by Withings is a smart mirror that provides AI-driven insights and tracks health metrics. The Omnia Smart Mirror also acts as a smart scale, heart rate monitor, and AI assistant in one, offering real-time health data directly from your reflection.

Why It Stands Out:

The Omnia Smart Mirror stands out by reimaging the mirror as a health tool. Similar to the Tonal workout station, where personal health metrics are clearly displayed. Adding to the appeal of the Omnia Smart Mirror is the option to track weight, cardio, body composition, and sleep patterns…if it ever launches.

LeafyPod: The Self-Watering Planter that thinks for you

What it Does:

LeafyPod is an AI-powered, self-watering smart planter that makes plant care effortless.

The LeafyPod is equipped with sensors that monitor soil moisture, light, temperature, and humidity, and it automatically adjusts watering schedules to suit your plant’s needs.

Why It Stands Out:

By automating plant care, LeafyPod will appeal to those who want green spaces but lack a green thumb. It ensures plants receive optimal care without constant attention.

The LeafyPod’s water reservoir can hold enough water to last up to four weeks, and a mobile app lets users monitor their plants and the surrounding environment.

AFEELA by Sony Honda Mobility: The intelligent EV

What It Does:

A collaboration between Sony and Honda, the Afeela is an electric car that blends advanced AI and sensor technology to elevate the driving experience.

The Afeela comes with 40 sensors, including cameras, LiDAR, radar, and ultrasonic units—Afeela offers automated driving assistance and immersive in-car entertainment.

Why It Stands Out:

One of its most unique features is in the cabin, where the driver can control in-car functions using natural voice prompts with the Afeela “Personal Agent” and receive activity suggestions.

Views and maps on the onboard display use Epic Games’ Unreal Engine, which hints at future features that could see the Afeela becoming not only a driving experience but also an entertainment hub.

While this is only a small sample of the innovations being unveiled at CES, it shows the AI arms race is still very much alive and well.

Edited by Sebastian Sinclair