A Chinese AI model called Kimi K3 has matched and even surpassed some leading American AI models while costing less. This challenges the idea that the U.S. is far ahead in advanced AI and may change the global AI competition.
Key Facts
Kimi K3 is a new AI model developed by Beijing-based Moonshot AI.
It performed better than top U.S. AI models like Anthropic's Fable 5 and OpenAI's GPT-5.6 in coding tests.
Kimi costs about 40% less than comparable U.S. AI models.
Moonshot plans to release Kimi as an open-weight model on July 27, letting users customize and run it themselves.
Earlier estimates said China was 6 to 12 months behind the U.S. in AI development, but Kimi shows the gap is closing fast.
Chinese AI labs may be using large amounts of American AI data to train their models.
China has obtained some restricted Nvidia chips through smuggling, despite U.S. efforts to limit access.
U.S. AI companies are developing newer models like GPT 6 and Claude Opus 5 to regain lead.
The U.S. government faces tough decisions balancing AI safety regulation and maintaining global competitiveness.
Read the Original
Want the full story? Tap a source to open the original
article.
A new study from the University of Virginia found that an unhealthy gut can cause bile acids to build up, which may help breast cancer spread in the body. Researchers hope this discovery will lead to new ways to prevent breast cancer from spreading and improve patient survival.
Key Facts
The study focused on hormone receptor-positive (HR+) breast cancer, the most common type that spreads to other parts of the body.
When the gut microbiome is unbalanced, bile acid levels become uncontrolled, affecting metabolism and the immune system.
An unhealthy gut environment helps breast cancer tumors spread more easily.
There are about 225,000 new HR+ breast cancer cases yearly in American women.
The researchers suggest that medications called bile acid sequestrants, already approved for metabolic diseases, might reduce cancer spread.
Higher bile acid levels and insulin resistance were linked to lower survival rates in HR+ breast cancer patients.
Some patients taking bile acid-suppressing drugs lived longer, but more research is needed to confirm the benefits.
The team is studying if these drugs can prevent cancer spread and exploring immune system effects related to bile acids.
Read the Original
Want the full story? Tap a source to open the original
article.
Many fans traveled long distances to watch Christopher Nolan’s new film, The Odyssey, at the BFI Imax cinema in London. The movie is the first to be shot entirely on the highest-quality Imax 1570 film format, using large, specialized cameras.
Key Facts
The Odyssey is based on Homer’s epic poem about Odysseus’s journey.
Fans came from places like Atlanta, Switzerland, and Ireland to see the film in London.
The film was shot entirely with Imax 1570 cameras, the highest-resolution format available.
The large cameras each weigh 180kg, and special soundproofing was developed to record dialogue.
The movie has a budget of about $250 million (£186 million).
It is expected to earn $80 million to $100 million in North America during its opening weekend.
Early reviews are very positive, with a 96-99% rating on Rotten Tomatoes.
Fans express admiration for Nolan’s filmmaking and have high expectations for The Odyssey.
Read the Original
Want the full story? Tap a source to open the original
article.
Clothing designers are creating garments with special patterns that aim to confuse facial recognition systems used in public spaces. These "adversarial clothing" designs use shapes and colors to make it harder for computer algorithms to identify people, raising awareness about privacy and surveillance concerns.
Key Facts
Facial recognition technology is increasingly used in public areas across Britain.
“Adversarial clothing” uses patterns that try to trick facial recognition software.
Advances in computing have made it easier to produce such garments for everyday consumers.
The effectiveness of these patterns depends on the type of surveillance system and the settings.
Experts warn that no design can fully guarantee protection from facial recognition.
Some garments incorporate features like infrared LEDs to confuse night-vision cameras.
There are concerns that facial recognition often misidentifies Black and Asian people more than white people.
Public worry about surveillance is growing, with nearly 60% fearing it may lead to a “surveillance society.”
Read the Original
Want the full story? Tap a source to open the original
article.
Scientists have discovered how a toxin made by a common gut bacterium called Bacteroides fragilis attaches to a protein on colon cells, causing damage that can lead to colorectal cancer. This breakthrough helps explain how some gut bacteria may directly contribute to the development of colon tumors and offers a new way to potentially block the toxin’s harmful effects.
Key Facts
Bacteroides fragilis is a common bacterium found in about 20% of healthy people’s guts.
Only certain strains of this bacterium produce a toxin called BFT linked to colorectal cancer.
The toxin binds to a protein on colon cells called claudin-4 to start damaging the gut lining.
Damage from BFT triggers inflammation, which may lead to tumor growth in the colon.
Researchers used a gene-editing method (CRISPR) to find claudin-4 as the toxin’s attachment point.
Blocking claudin-4 prevented the toxin from harming colon cells in lab tests.
Scientists created a molecular "decoy" that mimics claudin-4 to intercept the toxin and stop damage in animal studies.
Maintaining a healthy diet and lifestyle can help support colon health and reduce cancer risk.
Read the Original
Want the full story? Tap a source to open the original
article.
Konami is bringing back the Castlevania game series with a new side-scroller called Belmont’s Curse, set to launch in October. The game features Trevor Belmont’s daughter, Rose, fighting demons in 1499 Paris, and combines classic gameplay with new features developed by the team behind the game Dead Cells.
Key Facts
Belmont’s Curse is the first new Castlevania game since 2014.
The game takes place 23 years after the events of Castlevania 3 in 1989.
Players control Rose Belmont, daughter of Trevor Belmont, in demon-filled 1499 Paris.
Konami partnered with the external studio Evil Empire to develop the game.
The game uses a whip with new moves, allowing players to swing and attack enemies dynamically.
Players gain experience points (XP) by defeating enemies, which improves attack and defense stats.
Defeated bosses are turned into tarot card spells (arcana) that can be upgraded through gameplay.
Belmont’s Curse aims to offer a balanced challenge that is more forgiving than some difficult games in the genre.
Read the Original
Want the full story? Tap a source to open the original
article.
Robert Laidlow’s new album called Reality Eaters features music inspired by science and technology ideas. It includes pieces that explore topics like Einstein’s theories, Newton’s laws, and the impact of AI on creativity. The album is performed by groups like the BBC Philharmonic and showcases experimental uses of electronics and AI.
Key Facts
Reality Eaters is Robert Laidlow’s debut album on the NMC label.
The album mixes classical music with scientific concepts such as Einstein’s field equations and Newton’s law of gravity.
The piece Warp is a 12-minute piano concerto with orchestral parts inspired by space-time distortion.
Gravity is played by the Piatti Quartet and references Newton’s universal law.
Silicon is a three-part composition that explores AI and human creativity through music.
Silicon’s movements include Mind, Body, and Soul, involving machine imitation, electronic effects, and AI interacting with the BBC Philharmonic.
The album recording features the BBC Philharmonic orchestra and Vimbayi Kaziboni.
The music uses both traditional instruments and new technologies like adaptive electronics and AI algorithms.
Read the Original
Want the full story? Tap a source to open the original
article.
Chinese President Xi Jinping called for global cooperation in developing artificial intelligence (AI) during the World Artificial Intelligence Conference in Shanghai. He emphasized that AI should benefit people and remain under human control while warning against one country dominating the technology.
Key Facts
President Xi said AI development should involve many countries, not just one.
He highlighted the need for laws and systems to keep AI under human control.
China plans to work with countries in Africa, Latin America, Asia, and BRICS to share AI opportunities.
The US and EU have set limits on Chinese tech imports over security concerns.
China’s AI technology is improving and becoming cheaper, attracting users worldwide.
China is focusing on building its AI industry from chip making to everyday use through government support.
China produces large amounts of cheap electricity, which helps power big AI data centers.
China already produces more than twice the electricity of the US, supporting energy-heavy AI infrastructure.
Read the Original
Want the full story? Tap a source to open the original
article.
A podcast episode discusses the work of a philosopher who studies Google DeepMind's artificial intelligence (AI). The philosopher explores difficult questions about what AI really is and how it thinks.
Key Facts
The podcast features a philosopher working with Google DeepMind AI.
The philosopher tries to understand the nature of AI and its processes.
There is a discussion about the "deep mystery" of what AI truly is.
Google DeepMind is a company that develops advanced AI systems.
The episode aims to explain the challenges in understanding AI in simple terms.
Read the Original
Want the full story? Tap a source to open the original
article.
Telstra said a software problem with one of its key time-keeping servers caused a major mobile network outage across Australia. The server reset with the wrong date, which slowly spread errors in security systems, disrupting mobile calls and data for many customers.
Key Facts
The outage was caused by a software update that was not applied to a time-keeping server in Melbourne.
This server reset incorrectly to the year 2006 during maintenance.
The wrong date spread across the network, making security certificates invalid.
Customers had trouble connecting to the mobile network, affecting calls and data.
Telstra has three time servers in Melbourne, Sydney, and Perth, and the other two worked properly.
A design change made to fix an earlier problem was not documented, so maintenance staff were unaware of the server’s behavior.
Telstra took full responsibility and promised to improve controls to prevent future outages.
The emergency number 000 (triple zero) calls mostly worked, and fixed-line callers on the NBN were not affected.
Read the Original
Want the full story? Tap a source to open the original
article.
A new study by Meta’s Oversight Board found that major AI chatbots are less likely to criticize leaders in restrictive countries compared to those in freer countries. This shows that AI models may reflect government limits on free speech, potentially spreading these restrictions beyond national borders.
Key Facts
The study tested 10 popular AI chatbots, including those from Meta, Anthropic, and OpenAI.
Chatbots would create criticism of leaders like President Donald Trump or King Charles III but refused to criticize leaders of countries with strict speech laws.
Countries with speech restrictions included China, Saudi Arabia, Thailand, Cambodia, and Turkey.
The AI systems showed fewer critical responses about governments where expressing criticism is legally punished.
AI models may be biased due to training data influenced by government censorship or company decisions.
This bias could make it harder for people in free countries to use AI tools to support protests or political speech against restrictive regimes.
Another study found AI models trained on non-English data were also influenced by foreign government controls.
These findings raise concerns about AI’s role in spreading government censorship worldwide.
Read the Original
Want the full story? Tap a source to open the original
article.
SpaceX canceled the planned launch of its large Starship rocket and Super Heavy booster after some engines on the booster failed to start during ignition. Engineers will replace two engines, and the company expects to try launching again early next week.
Key Facts
The launch took place at SpaceX’s Starbase in Texas near the US-Mexico border.
The rocket is over 400 feet tall and uses liquid methane and oxygen as fuel.
The Super Heavy booster has 33 Raptor engines that provide the thrust for liftoff.
Four of the 33 engines did not ignite, causing an automatic shutdown of the launch.
This was the 13th full-scale Starship launch attempt and the second using the latest Raptor 3 engines.
SpaceX plans to replace two engines before the next launch attempt.
A previous test flight in May faced engine issues during flight and landing, which SpaceX aims to fix.
The company wants to achieve a controlled landing and reuse of the booster on future flights.
Read the Original
Want the full story? Tap a source to open the original
article.
Scientists have created tiny particles called nanoparticles that use copper already inside tumors to kill cancer cells. These nanoparticles deliver a special chemical directly to cancer cells to trigger a type of cell death called cuproptosis, while minimizing harm to healthy cells.
Key Facts
Researchers designed biodegradable nanoparticles made from a safe material called PLGA-PEG.
The nanoparticles have a surface modified with iRGD, a peptide that helps them target and enter cancer cells.
They carry a chemical called TPEN which binds copper inside cancer cells to cause cell death.
The nanoparticles are about 80 nanometers in size and release TPEN slowly over 72 hours.
Experiments showed these particles entered breast cancer cells more effectively than non-targeted ones.
Targeted nanoparticles killed cancer cells better and were less harmful to normal cells compared to free TPEN.
The approach uses copper naturally present in tumors instead of adding extra copper, aiming to reduce side effects.
Although promising, more research is needed to ensure safety and effectiveness before it can be used in patients.
Read the Original
Want the full story? Tap a source to open the original
article.
Chinese AI company Moonshot AI released a large new AI model called Kimi K3, which may match top American AI systems but costs less. Early tests show Kimi K3 performing better than some leading U.S. AI models in coding and text tasks, raising concerns about U.S. leadership in artificial intelligence.
Key Facts
Kimi K3 contains 2.8 trillion parameters, making it one of the largest open-weight AI models.
It can handle very large amounts of text at once (1 million-token context window) and works with both text and images.
In blind tests, Kimi K3 outperformed leading U.S. models like Anthropic’s Fable 5 and OpenAI’s GPT-5.6 Sol in coding tasks.
Kimi K3 also ranked higher than Anthropic’s Opus 4.8, a recent top AI model, in broader text tests.
Moonshot AI is offering Kimi K3 at prices significantly lower than top U.S. AI models.
The model’s "weights" (details allowing others to run or adjust it) will be released on July 27, after which outside developers can inspect it.
Chinese AI progress is seen as a growing challenge to U.S. dominance in advanced AI technology.
The launch happens ahead of the 2026 World Artificial Intelligence Conference in Shanghai, where China will discuss AI plans.
Read the Original
Want the full story? Tap a source to open the original
article.
New York has become the first state to temporarily stop building new data centers. Environmental advocate Erin Brockovich talked about the importance of this decision on a CBS News program called "The Takeout."
Key Facts
New York state has placed a temporary ban on data center construction.
This ban is the first of its kind in the United States.
Data centers are facilities that house computer servers and related equipment.
Environmental concerns are a main reason for stopping new data center projects.
Erin Brockovich, known for her work on environmental issues, spoke about this topic on CBS News.
Other states might adopt similar bans following New York’s example.
The discussion took place on a CBS News segment named "The Takeout."
Read the Original
Want the full story? Tap a source to open the original
article.
People are worried about how children and teenagers use artificial intelligence (AI) every day. Some critics say tech companies are not doing enough to keep kids safe and want stronger rules. The Omidyar Network, a group that supports tech projects, has proposed a new way to help protect young users.
Key Facts
There are growing concerns about children and teens using AI daily.
Critics believe tech companies are not sufficiently protecting young users.
Some people argue for more government rules to regulate AI for kids.
The Omidyar Network is a tech-focused philanthropic organization.
CEO Michele Jawando from the Omidyar Network shared a new idea to improve safety for children using AI.
The discussion was highlighted by journalist Geoff Bennett.
Read the Original
Want the full story? Tap a source to open the original
article.
Thousands of Google workers have signed a petition asking the company to protect employees from layoffs as it invests heavily in artificial intelligence (AI). The petition requests better severance deals, alternatives to layoffs, and changes to performance reviews. Google recently cut jobs even while expanding its AI efforts, and other tech companies are also reducing staff partly because of AI.
Key Facts
Over 4,500 Google employees signed a petition for layoff protections and better severance.
The petition asks for buyouts before mandatory layoffs and the option to use severance as paid leave.
Workers want to end performance ratings that focus on quotas instead of merit.
Google’s valuation has grown to $4 trillion, but it is still cutting jobs, especially in management roles.
The petition was delivered to CEO Sundar Pichai’s office but got little direct response.
Other tech firms like Meta, Oracle, Block, and Microsoft have also laid off workers, citing AI-related reasons.
Meta employees sued their company alleging AI tools were used to target them for layoffs after taking protected leave.
Alphabet’s CFO said AI is a key investment, and the company plans to continue hiring in this area.
Read the Original
Want the full story? Tap a source to open the original
article.
T-Mobile made a technical mistake while moving some customers to new phone plans, resulting in the loss of free phone lines that many longtime users had. The company said it is aware of the errors and is working quickly to fix the problem and restore any lost discounts.
Key Facts
T-Mobile forced many customers to switch from older plans to new rate plans.
Some customers lost free phone lines they had earned through previous promotions during this switch.
Losing these free lines caused some bills to rise sharply, sometimes by hundreds of dollars.
T-Mobile admitted the problem was due to a technical error in applying promotional discounts.
The company is fixing the issue by restoring discounts and backdating them where needed.
Other billing problems, such as incorrect charges for Hulu, are also being investigated.
T-Mobile promised customers would keep their benefits and said improvements would come with the new plans.
Free lines were originally offered as promotions for longtime customers with multiple paid lines.
Read the Original
Want the full story? Tap a source to open the original
article.
Anthropic's CEO, Dario Amodei, donated $1 million to Public First, a political group that supports candidates favoring stricter rules on artificial intelligence (AI). Several other Anthropic employees also contributed a total of $2.15 million during the last three months.
Key Facts
Dario Amodei is the CEO of Anthropic, an AI company.
He gave $1 million to Public First, a super PAC.
Public First supports candidates who want stronger AI regulations.
Other Anthropic employees donated $2.15 million together.
The donations were reported in a campaign finance filing on Wednesday.
A Google DeepMind engineer also made a contribution, according to the article.
The donations were made during the last quarter (three months).
Read the Original
Want the full story? Tap a source to open the original
article.
The European Commission has enforced new rules under the Digital Markets Act that require Google to share search data with competitors and allow other AI services equal access on Android devices. These rules aim to increase competition and user choice in the EU but Google warns the changes might affect privacy and security.
Key Facts
The Digital Markets Act (DMA) is a new European law targeting big tech companies like Google, Apple, and Meta to promote competition.
Google must now let other AI assistants access Android features equally, not just its own Gemini AI assistant.
Google currently preloads Gemini AI on Android phones and gives it special permissions like responding to "Hey Google."
The European Commission wants users to install and use any AI assistant on Android without losing functionality.
Google has to share search data with other search companies for a reasonable fee to help competitors grow.
The Commission calls this data sharing necessary because previous Google offers did not meet competition requirements.
Google argues these measures could reduce privacy and security protections for users.
The EU says it designed the rules to protect user privacy while boosting competition and is willing to tweak them if needed.
Read the Original
Want the full story? Tap a source to open the original
article.
