July 25, 2026

AI News (07/25) : South Korea Seals $1.23 Trillion in Deals as Tech Giants Push for Next-Gen AI Chips

The global landscape of artificial intelligence is shifting under the weight of trillion-dollar hardware alliances, massive local compute demands, and increasingly fierce model-pricing battles. From South Korea’s historic chip summits to critical open-source debates and medical AI integration, today’s developments are rewriting the rules of technology.

1. Samsung, SK Group seal $1.23 trillion in deals as South Korea hosts AI powers

At a high-profile summit in San Francisco, South Korean President Lee Jae Myung unveiled the "San Francisco AI Declaration," marking $950 billion ($1.23 trillion SGD) in massive hardware initiatives. SK Group finalized deals worth $750 billion, including a $500 billion-plus partnership with Nvidia to construct massive data centers and secure high-bandwidth memory. Simultaneously, Samsung Electronics signed a $200 billion semiconductor agreement with Broadcom to provide cutting-edge packaging and sub-2-nanometer foundry services.

Why it matters: As the global shortage of high-speed processing chips threatens to stall AI development, tech companies are locking in physical hardware pipelines directly with South Korea's dominant memory and foundry leaders.

2. Prime Video's upcoming AI makeover could completely change how you decide what to watch next

Amazon is developing "Project Lighthouse," a major Prime Video redesign personally championed by founder Jeff Bezos. After rejecting earlier, less ambitious proposals, Bezos pushed for a fresh, generative-AI-driven approach under design head Kam Keshmiri. Backed by a portion of Amazon’s massive $200 billion AI capital expenditures, the overhaul will introduce AI-curated content tiles and voice-controlled search to improve personalization for over 200 million active viewers.

Why it matters: Streaming platforms are transitioning from basic algorithms to deeply integrated generative AI to boost viewer engagement in an increasingly crowded media market.

3. Anthropic's Claude Opus 5 delivers near-Fable 5 performance at half the token price

Anthropic has launched Claude Opus 5, its new flagship model built to challenge OpenAI's GPT-5.6 Sol and Fable 5. Keeping rates at $5 per million input tokens and $25 per million output tokens, Opus 5 introduces five user-selectable effort settings to manage costs and latency, alongside a 2.5x Fast Mode. While scoring exceptionally well on agentic coding and knowledge benchmarks, the model's highest effort level has shown some benchmark quirks due to excessive code-refactoring behaviors.

Why it matters: Anthropic is aggressively driving down operational costs to remain competitive against rival models while giving developers unprecedented control over their cost-to-performance ratios.

4. Waymo’s driverless cars involved in 68% fewer crashes than human drivers, U.S. study finds

A three-year study by the Insurance Institute for Highway Safety across four major U.S. cities revealed that Waymo autonomous vehicles were involved in 68% fewer crashes than human drivers. Compiling over 80 million driverless kilometers, the study logged a crash rate of 1.28 per million miles for Waymo compared to 4.06 for human drivers. Furthermore, Waymo’s crashes tended to be significantly less severe, avoiding dangerous head-on or rear-end collisions.

Why it matters: Independent, empirical safety data remains the most critical tool for autonomous vehicle companies seeking to ease regulatory roadblocks and secure public trust.

5. Voters Are Being Inundated by a Barrage of AI-Generated Election Ads

In local and congressional elections across Missouri, California, and Kentucky, political campaigns are being flooded with sophisticated AI-generated attack ads. In Missouri, conservative candidate Jason Law utilized cloned voices to depict rivals in a tax conspiracy, while PAC-funded deepfakes in Kentucky showed fake security footage of Representative Thomas Massie. The high-tech smear campaigns appear effective, with several AI-targeted incumbents losing their races.

Why it matters: Generative AI tools have lowered the cost of creating convincing political propaganda, raising urgent concerns about election integrity and voter manipulation.

6. ChatGPT now has a space for sharing medical records. Should you?

OpenAI has launched "Health in ChatGPT," allowing its 300 million weekly health-seeking users to upload lab results, medical records, and sync Apple Health data. However, medical and AI experts warn of severe risks, including hallucinated medical advice. The launch comes amidst ongoing lawsuits and tragic incidents where users suffered critical health issues or death after acting on ChatGPT's health guidance.

Why it matters: Integrating personal health data into general-purpose chatbots democratizes healthcare advice, but creates immense liability risks and potential dangers for vulnerable users.

7. Demand for AI Computing Is So High That Apple Is Redesigning Every Mac It Sells

Apple is overhauling its entire Mac lineup to keep up with developers and hobbyists running AI agents locally. The redesign includes touchscreen and OLED MacBook Pros alongside refreshed iMacs. Extreme demand for local AI hardware and ongoing memory shortages have already pushed back shipping times for Mac minis and Mac Studios by up to three months.

Why it matters: The shift toward executing heavy AI models locally rather than in the cloud is forcing hardware giants to completely re-engineer consumer hardware for high-performance compute.

8. China's cleaning robots mop up 70% of global share through innovation

Chinese smart-home manufacturers now command 70% of the global market for household cleaning robots. Companies like Roborock are maintaining this dominance by focusing on unique engineering breakthroughs—such as the stair-climbing Saros Rover—rather than competing purely on cheap manufacturing costs.

Why it matters: Chinese hardware firms are securing global markets by out-innovating Western competitors in physical, embodied AI applications.

9. OpenAI quietly signs letter from Nvidia, Microsoft, and Meta warning about dangers of 'premature restrictions' on open-weight AI models

OpenAI has joined Nvidia, Meta, Microsoft, and others in signing an open letter defending open-weight AI models against premature government regulations. The unified front arrives as geopolitical tensions flare, with the US Treasury accusing Chinese firm Moonshot AI of conducting "industrial-scale distillation attacks" on Anthropic's Claude Fable 5 to build its new open-weight Kimi K3 model.

Why it matters: The open-source AI debate has escalated into a national security conflict, forcing tech leaders to defend open innovation while governments threaten sanctions over intellectual property theft.

10. Midjourney announces surprise purchase of popular horoscope app Co-Star

AI art powerhouse Midjourney has acquired the popular astrology app Co-Star, with founder Banu Guler joining Midjourney as Chief Design Officer. The unusual purchase is part of Midjourney’s broader expansion into holistic health and wellness, which includes the development of diagnostic ultrasonic scanners and physical spa concepts.

Why it matters: Generative AI leaders are diversifying into wellness, lifestyle, and consumer design to embed their technology into daily human habits beyond digital image generation.

To stay up to date on everything going on in AI, check out the tracker at the500feed.com

Sources

View the live feed — updated daily from 500+ sources →← All posts