The Rundown

The Rundown Get the latest in AI in 5 minutes a day. Read by 1,000,000+ early adopters.

OpenAI CEO Sam Altman has cautioned users against treating ChatGPT as a therapist, warning that conversations with the A...
07/25/2025

OpenAI CEO Sam Altman has cautioned users against treating ChatGPT as a therapist, warning that conversations with the AI lack legal confidentiality.

Speaking on Theo Von’s podcast, Altman noted that while people often share deeply personal information with ChatGPT, those interactions aren’t protected by doctor-patient or therapist-client privilege.

This creates a significant privacy risk, as OpenAI could be legally compelled to hand over user chats in court. The issue highlights the absence of a legal framework for AI communications and raises concerns over user privacy, especially as ChatGPT is increasingly used for emotional support and life advice by young people.

Source: Theo Von podcast

Walmart has unveiled four AI-powered “super agents” aimed at transforming retail operations and intensifying its competi...
07/25/2025

Walmart has unveiled four AI-powered “super agents” aimed at transforming retail operations and intensifying its competition with Amazon.

The agents:

Sparky (customer-facing)
Associate (employee support)
Marty (supplier automation)
Developer (AI testing)

are designed to operate with minimal human oversight, executing complex tasks across Walmart’s vast ecosystem.

Sparky, already active in the app, will now handle reorders, party planning, and fridge-based recipe suggestions. Walmart’s goal is for e-commerce to comprise 50% of its revenue within five years.

As part of the push, Walmart hired ex-Instacart CPO Daniel Danker as EVP of AI Acceleration, signaling a bold new era in AI-led retail.

Source: Reuters
Image: Getty Images

A federal judge has certified a class action lawsuit against AI company Anthropic, exposing it to potential damages of u...
07/25/2025

A federal judge has certified a class action lawsuit against AI company Anthropic, exposing it to potential damages of up to $1.05 trillion.

The case alleges Anthropic pirated over 6 million copyrighted books from shadow libraries like LibGen and PiLiMi to train its AI model Claude. While the court previously ruled that using such material for training may fall under “fair use,” it found storing the books to build a research library crossed into copyright infringement.

Legal experts warn this case could pose an existential threat to Anthropic and set a precedent that forces AI firms to license training data.

Source: Los Angeles Times

Scientists at the Technical University of Denmark have created an AI-powered platform that designs cancer-targeting prot...
07/25/2025

Scientists at the Technical University of Denmark have created an AI-powered platform that designs cancer-targeting proteins in weeks instead of years.

It uses three different AI tools, including one from Google, to create tiny proteins that help immune cells find and kill cancer cells like melanoma. The system can make custom proteins for both common and unique cancer types, and it checks designs to make sure they won’t harm healthy cells.

This could speed up drug development and open the door to personalized cancer treatments—where AI creates medicine specially made for each patient’s needs.

Source: Science.org

OpenAI is reportedly planning to launch its next-gen GPT-5 in August, according to a report from The Verge — with CEO Sa...
07/25/2025

OpenAI is reportedly planning to launch its next-gen GPT-5 in August, according to a report from The Verge — with CEO Sam Altman also speaking about the new model this week in a conversation with comedian Theo Von.

OpenAI is set to unveil GPT-5 in August, according to The Verge, with CEO Sam Altman confirming its arrival during a con...
07/25/2025

OpenAI is set to unveil GPT-5 in August, according to The Verge, with CEO Sam Altman confirming its arrival during a conversation with comedian Theo Von.

GPT‑5 will merge advanced language skills with o3-style reasoning, creating a unified model for a broader range of tasks. Altman described early tests as “a here it is moment,” highlighting the model’s impressive capabilities.

Additionally, OpenAI plans to release its first open-weight model in years by the end of July. Both releases are expected to mark a major leap forward in AI performance and accessibility.

Nvidia AI chips worth over $1 billion were reportedly smuggled into China in the three months following tightened U.S. e...
07/24/2025

Nvidia AI chips worth over $1 billion were reportedly smuggled into China in the three months following tightened U.S. export controls, according to the Financial Times.

The restricted B200, H100, and H200 processors—banned under Trump-era policies—have appeared on China’s black market, with distributors from Guangdong, Zhejiang, and Anhui selling them to AI data center suppliers. Southeast Asian countries reportedly served as transit points.

Nvidia warned that using unauthorized chips limits technical support and efficiency. While the U.S. has eased some restrictions recently, further controls—possibly affecting nations like Thailand—are being considered by the Commerce Department as early as September.

Source: Financial Times

OpenAI is reportedly preparing to launch its next-generation GPT-5 model in early August 2025. CEO Sam Altman has hinted...
07/24/2025

OpenAI is reportedly preparing to launch its next-generation GPT-5 model in early August 2025. CEO Sam Altman has hinted at a summer release, emphasizing safety and alignment.

GPT-5 is expected to feature unified multimodal capabilities—handling text, image, audio, and video—alongside improved reasoning, a context window exceeding one million tokens, and reduced hallucinations.

It will launch with mini and nano versions and offer API access for developers. Some enterprise clients have already received early demos. Unlike past releases, the rollout has been low-profile, with updates shared through interviews and social media rather than official press events.

Source: The Verge

Japanese shipping giant Mitsui O.S.K. Lines (MOL) and Turkish energy firm Karpowership's subsidiary Kinetics have announ...
07/24/2025

Japanese shipping giant Mitsui O.S.K. Lines (MOL) and Turkish energy firm Karpowership's subsidiary Kinetics have announced plans to launch a 73-megawatt floating data center by 2027.

Housed on a 120-meter-long ship, the data center will be powered by liquefied natural gas and renewable energy sources like offshore wind and onshore solar. The project aims to rapidly deploy digital infrastructure while easing strain on land-based power grids—a growing concern amid rising AI demand.

The floating design enables faster deployment and global mobility. MOL sees this as a flexible, lower-cost interim solution as traditional data center builds face delays and grid bottlenecks.

Source: Tom’s Hardware
Image: Karpowership

Yahoo Japan has announced a bold company-wide policy requiring all 11,000 employees to use generative AI tools in their ...
07/24/2025

Yahoo Japan has announced a bold company-wide policy requiring all 11,000 employees to use generative AI tools in their daily work. The move is part of a broader strategy to double productivity by 2028.

Starting with automating 30% of routine tasks—like meeting notes, expense processing, and research—the company is deploying internal tools such as SeekAI to streamline operations. Rather than replacing workers, Yahoo Japan aims to free up time for higher-level thinking and decision-making.

The initiative reflects a growing global trend of embedding AI into everyday workflows to enhance, not displace, human roles—marking a shift in how enterprises adopt AI.

Source: Techradar

YouTube has released new AI tools for Shorts to help creators make videos more easily. One tool turns photos into short ...
07/24/2025

YouTube has released new AI tools for Shorts to help creators make videos more easily. One tool turns photos into short videos using Google’s Veo 2 model.

Another, called Generative Effects, lets users change selfies or drawings into fun, AI-made visuals. These features are available in the Shorts camera and a new section called AI Playground, where users can also create videos, images, and music.

YouTube says all AI-made content will have labels and watermarks to show it was made with AI. An upgrade to Veo 3 is expected soon.

Source: Google

Address

Vancouver, BC

Website

Alerts

Be the first to know and let us send you an email when The Rundown posts news and promotions. Your email address will not be used for any other purpose, and you can unsubscribe at any time.

Contact The Business

Send a message to The Rundown:

Shortcuts

Share

Category