Chinese AI: Tensions in Washington and Copyright Records

Chinese AI Models Plunge Trump's AI World into Discord

Last weekend was marked by a series of tensions within the Trump administration regarding the management of Chinese artificial intelligence models. Current and former advisors to the president exchanged heated remarks about major American AI companies. David Sacks, for instance, criticized the models developed by Anthropic, calling them "lobotomized" and "woke." Meanwhile, Emil Michael, a senior official at the Pentagon, referred to the new head of future strategies at OpenAI as a "supreme village idiot."

This situation has been exacerbated by the lack of consensus on how to deal with Kimi, a free and open-source AI model recently launched by the Chinese company Moonshot. This model appears to compete with the offerings from OpenAI and Anthropic, which are not free. Each launch of a free model in China diminishes the appeal for American companies to pay for models from Anthropic or OpenAI, thus creating economic and political tensions for the president and dividing his top AI advisors.

Summary of Key News

1. Anthropic's record $1.5 billion settlement approved

   Anthropic has agreed to pay a record amount of $1.5 billion to settle allegations that it used pirated works to train its AI model, Claude. This settlement is the largest ever recorded for copyright issues, but despite this, many authors and creators do not view the decision as a victory. Growing concerns around AI copyright could stifle creativity in the sector.

2. Trump administration considers a ban on Chinese AI models

   The launch of the Kimi K3 model by Moonshot has reignited discussions about the need to restrict access to Chinese AI models. However, within the administration, opinions differ on the best approach to take. China's open-source strategy seems to be paying off, complicating matters further for American decision-makers.

3. China considers stricter export controls on AI models and chips

   Beijing is contemplating tightening its export controls to prevent the West from accessing its advanced technologies in AI and chips. Discussions have taken place between the Chinese government and local tech companies to consider potential restrictions.

4. Trump's AI security chief resigns after just three months

   Chris Fall, who led the Federal AI Security Institute (CAISI), has left his position after only three months in office. No official reason has been provided for his departure.

5. Google working on a new chip to run Gemini models more efficiently

   Google is developing a new chip, named "Frozen V2," which could be deployed by 2028 to enhance the efficiency of Gemini models. This announcement led to a rise in Alphabet's stock.

6. New Orleans police explored arming drones with weapons

   A draft manual for drones has been revealed, paving the way for the use of armed quadcopters to combat shoplifters, according to recent reports.

7. EU fines AliExpress a record amount for selling dangerous products

   AliExpress has been fined €550 million, the largest ever imposed under the Digital Services Act, for selling products deemed hazardous. Alibaba has announced its intention to appeal this decision.

8. AI chatbot electoral advice is "inaccurate and unreliable"

   Trials conducted in Hungary concluded that electoral advice provided by AI chatbots lacks accuracy and reliability.

9. Red light therapy shows promise for healing and healthy aging

   Recent studies suggest that red light therapy could improve skin health and reduce vision loss, offering interesting prospects for healthy aging.

10. Neill Blomkamp's new horror clip is entirely AI-generated—and it's bad

    Director Neill Blomkamp experimented with an entirely AI-generated horror clip, receiving negative reviews. Nonetheless, he is considering making a full-length feature in this format.

Quote of the Day

"It would be a terribly counterproductive form of intervention if that were to happen."
— Tech investor Chamath Palihapitiya critiques plans to restrict Chinese AI models in a post on X.

Inside Chicago's Surveillance Panopticon

On the morning of September 2, 2024, a tragic incident occurred in Chicago where four people were shot on a westbound train. The police quickly activated their digital surveillance network, a system that connects thousands of cameras throughout the city, allowing for the suspect's arrest just 90 minutes after the incident.

Authorities claim that this surveillance system is essential for maintaining public safety. However, activists and many residents are concerned about the chilling effect of this "surveillance panopticon" on behavior and potential violations of privacy and free speech rights.

We Can Still Have Nice Things

A space for comfort, pleasure, and distraction to brighten your day. (Have ideas? Write to me.)

• NASA shared a stunning timelapse video of Mars as seen by the Psyche spacecraft.
• This comparison of American and European urban planning shows that good urban design is a choice.
• Musician Luca Stricagnoli recently performed a wonderful acoustic guitar medley of Prodigy songs.
• Two Australian paddleboarders rescued a stranded wallaby after it was swept out to sea—and filmed the entire rescue.