Scale AI warns China as Apple partners with Google on Gemini

Key Takeaways

• The Trump administration accuses China of running industrial-scale campaigns to steal U.S. AI technology using proxy accounts and jailbreaking techniques.
• Michael Kratsios, the top science and technology advisor, stated these groups aim to strip security protocols from copied American AI models.
• Apple CEO Tim Cook is stepping down on September 1, leaving the company behind in the AI race compared to rivals like Google, Microsoft, and OpenAI.
• John Ternus will lead Apple starting September 1, 2026, with a new partnership with Google to base foundation models on Gemini technology.
• Meta plans to track employee keystrokes on Google, LinkedIn, and Wikipedia as part of its Model Capability Initiative to train AI models.
• An Oregon appeals court ruled that lawyers must disclose when generative AI creates errors in court filings and may face legal sanctions.
• A poll of 1,000 UK workers found high earners are more than twice as likely to use AI in the workplace as low earners.
• Five technology companies launched AI health tools in 2026, but studies show earlier models performed no better than users managing their own resources.
• Search engine optimization is shifting from keyword-first tactics to product-first discovery driven by structured data and informative product pages.
• UNC Charlotte launched an AI chatbot named Charlie to handle student IT issues, though some students expressed safety concerns about sharing passwords.
• Three large language model agents, including GPT-5.4 Pro, Gemini 3.1 Pro, and Claude Opus 4.6, secured first place in a Kaggle competition by generating over 600,000 lines of code.
• Nigerian filmmaker Malik Afegbua uses AI to record conversations with older generations to preserve cultural history and wisdom.

Trump Administration Accuses China of Industrial AI Theft

The Trump administration accused Chinese entities of running industrial-scale campaigns to steal U.S. artificial intelligence technology. Michael Kratsios, the top science and technology advisor, stated that these groups use tens of thousands of proxy accounts and jailbreaking techniques to extract capabilities from American AI models. The government plans to share this information with U.S. AI companies and explore measures to hold foreign actors accountable. While legitimate AI distillation is acceptable, these unauthorized attacks undermine American research and strip security protocols from copied models.

U.S. Vows Crackdown on Foreign AI Model Exploitation

The Trump administration issued a memo accusing foreign companies, mainly in China, of deliberately exploiting U.S. artificial intelligence models. Kratsios wrote that these campaigns aim to steal American expertise and innovation through industrial-scale distillation efforts. The administration will work with U.S. AI firms to identify such activities and build defenses against them. This action comes as China narrows the performance gap with the U.S. in the AI race, according to a recent Stanford University report. A new bipartisan bill also seeks to punish foreign actors who extract key technical features from closed-source American AI systems.

White House Memo Details China AI Distillation Threats

The White House Office of Science and Technology Policy released a memo detailing how Chinese entities are engaged in deliberate campaigns to distill U.S. frontier AI systems. The document explains that legitimate distillation creates smaller, cheaper models, but adversarial distillation involves competitors systematically querying models to harvest proprietary information. These campaigns use proxy accounts and jailbreaking to evade detection and strip security protocols from the resulting models. The administration plans to share intelligence with industry and develop best practices to mitigate these industrial-scale attacks.

FT Reports White House Accuses China of AI Theft

The Financial Times reported that the White House accused China of stealing U.S. artificial intelligence intellectual property on an industrial scale. Michael Kratsios shared a memo stating that foreign entities are leveraging proxy accounts and jailbreaking techniques to expose proprietary information from American AI models. The Chinese Embassy in Washington opposed the baseless allegations, claiming China protects intellectual property rights. The memo was released just weeks before a summit between President Donald Trump and Chinese leader Xi Jinping. The administration will share information with American companies to help them combat these distillation efforts.

U.S. Warns of Industrial AI Theft Campaigns

The U.S. government accused China of running industrial-scale campaigns to steal AI secrets from American companies. Michael Kratsios explained that these attacks involve querying proprietary models millions of times to build datasets that replicate system behavior. Such tactics allow foreign actors to release models that appear to match U.S. capabilities at a fraction of the cost. The warning comes as Trump prepares for a high-stakes meeting with China. Kratsios noted that while distilled models may not hold up over time, they enable the release of products that strip away security protocols and ideological neutrality.

White House Issues Warning Ahead of Trump-Xi Summit

The White House Office of Science and Technology Policy accused China of industrial-scale AI technology theft in a memo released just weeks before the Trump-Xi summit. The U.S. has evidence that foreign entities are running distillation campaigns to steal American AI technology and strip security protocols. The accusation precedes the first face-to-face meeting between the two leaders since 2025. Anthropic, the maker of the Claude chatbot, claimed China used a mass-proxy distillation process to siphon key data from their models. The Pentagon previously labeled this a supply chain risk for the U.S.

China Denies AI Theft Accusations as U.S. Prepares Crackdown

The U.S. is preparing to crack down on China's alleged industrial-scale theft of American artificial intelligence intellectual property. The Financial Times reported that Michael Kratsios warned foreign entities are using proxy accounts and jailbreaking to expose proprietary information. The Chinese embassy spokesperson called the White House accusations pure slander and emphasized China's commitment to protecting intellectual property rights. Congress is considering updating laws to equip U.S. companies fighting this alleged fraud. The memo comes ahead of Donald Trump's anticipated meeting with China's president Xi Jinping next month.

Tim Cook Legacy Marked by Missed AI Revolution

Apple CEO Tim Cook is stepping down on September 1 to become executive chairman, leaving a legacy that includes a missed opportunity in the AI revolution. Although Apple thrived under his leadership, the company was years behind in the AI race when ChatGPT emerged in late 2022. Cook's failure is attributed to Apple's bias toward doing everything in-house rather than partnering with external experts. The incoming CEO, John Ternus, must now play catch-up with rivals like Google, Microsoft, and OpenAI. Apple has announced a partnership with Google to base its next generation of foundation models on Google's Gemini technology.

John Ternus Leads Apple Through Major Product Launches

John Ternus will take over as Apple CEO on September 1, 2026, just days before the company's expected September event. The new lineup is expected to feature the iPhone 18 Pro, iPhone 18 Pro Max, and a new foldable device called the iPhone Fold. Apple also plans updates to the Apple Watch Series 12 and Ultra 2, along with a higher-end version of AirPods Pro with infrared cameras for visual intelligence. The smart home push includes a new hub with a display, an updated Apple TV 4K, and new HomePod models. Ternus's background as a hardware engineer suggests a stronger emphasis on devices as AI becomes central to the strategy.

Meta Tracks Employee Keystrokes for AI Training

Meta plans to track employee keystrokes on Google, LinkedIn, and Wikipedia as part of its Model Capability Initiative to train AI models. The internal tool allows Meta to observe staffers' actions on the web, including mouse movements and button clicks, to gather real examples of how people use computers. Meta properties like Threads and Manus are also on the list of sites being monitored. A spokesperson stated that safeguards are in place to protect sensitive content and that the data is not used for any other purpose. Some employees characterized the project as dystopian due to concerns about exposing passwords and personal information.

Appeals Court Orders Lawyers to Disclose AI Errors

An Oregon appeals court ruled that lawyers must be candid with judges when using generative AI for legal work creates errors in court filings. The court imposed legal fees as a sanction for a brief containing fabricated quotations and other errors created with AI assistance. Attorney Abby Shearer acknowledged using generative AI as a research and drafting aid but did not identify which specific tools she used. The court stated that regardless of the provider, a generative AI program is not itself a lawyer. The plaintiff's attorneys were ordered to pay about $8,000 in legal fees to the defendant.

AI Use in Workplace Widens Gap Between High and Low Earners

A new poll found that high earners are racing ahead with AI use while the workplace divide widens. The survey of 1,000 UK workers showed that those in the top 10 percent of earners were more than twice as likely to use AI as those in the bottom 10 percent. Men were also 50 percent more likely to use AI than women, and workers in London were 30 percent more likely to use it than those elsewhere. The most common uses of AI in the workplace were for data analysis, customer service, and marketing. Lower-paid workers are less likely to have access to AI training, which could exacerbate existing inequalities in the job market.

Tech Firms Launch AI Health Tools Despite Accuracy Concerns

Five technology companies have launched dedicated consumer-facing AI health tools in 2026, though questions about their accuracy remain. These applications allow users to connect medical records and wearable data to receive personalized health guidance. A study found that participants using earlier AI chatbot models performed no better than a control group using their own resources for medical scenarios. Users sometimes received conflicting advice because the chatbots misinterpreted prompts or gave inconsistent responses. While newer models may provide higher performance on benchmarks, it remains unclear if these improvements translate to real-world gains.

Product Pages Become Key for SEO and AI Discovery

Search engine optimization is shifting from keyword-first tactics to product-first discovery driven by product pages and structured data. Technical SEO still matters, but its value is now measured by clarity and decision support across the user journey rather than just rankings. Implementing schema markup, designing informative product descriptions, and optimizing site architecture reduce friction for users and AI models. Product teams that integrate SEO into product design reduce misclassification and improve conversion signals. This approach treats product content as a durable asset that helps search engines and recommendation systems surface products accurately.

UNC Charlotte Launches AI Chatbot for Student Tech Support

UNC Charlotte introduced an AI chatbot named Charlie to handle student information technology issues and free up staff for larger tasks. The chatbot summarizes information from analytics and databases to answer questions about passwords, usernames, and other common IT problems. Students can access Charlie by searching for the Office of OneIT link and chatting during business hours from 7:30 a.m. to 9 p.m. Monday through Thursday. Outside of business hours, students can submit help tickets for issues to be answered when the desk opens again. Some students expressed safety concerns about sharing personal information like passwords with an AI chatbot.

Generative AI Agents Help Team Win Kaggle Competition

Three large language model agents worked together to generate over 600,000 lines of code and secure first place in a Kaggle competition. The team used GPU acceleration from NVIDIA tools to run 850 experiments and quickly iterate on machine learning models. The workflow guided the agents through exploratory data analysis, building baseline models, and performing feature engineering. The solution combined multiple models using a human-in-the-loop approach with GPT-5.4 Pro, Gemini 3.1 Pro, and Claude Opus 4.6. This method dramatically compressed the loop of generating, testing, and improving ideas in modern machine learning competitions.

Nigerian Filmmaker Uses AI to Preserve History

Malik Afegbua is a Nigerian filmmaker using artificial intelligence to preserve history for future generations. He is recording conversations with older generations to capture cultural knowledge and stories. NBC News reported on this project which aims to document history through the use of AI technology. The initiative seeks to ensure that the experiences and wisdom of older generations are not lost to time. Afegbua's work highlights a creative application of AI beyond technology and business to support cultural preservation efforts.

Sources

• White House warns of 'industrial-scale' efforts in China to rip off U.S. AI tech
• Trump administration vows crackdown on Chinese companies 'exploiting' AI models made in US
• US Govt Says Chinese Entities Are Distilling US AI Models, Will Explore Measures To Hold Foreign Actors Accountable
• White House accuses China of ‘industrial scale’ theft of AI technology, FT reports
• U.S. accuses China of "industrial-scale" campaigns to steal AI secrets
• White House accuses China of 'industrial-scale' AI technology theft weeks ahead of Trump-Xi summit
• US accuses China of “industrial-scale” AI theft. China says it’s “slander.”
• Tim Cook’s legacy: a successful CEO who stumbled over AI
• John Ternus Takes Over Apple as New iPhones, Other AI Products Near
• Meta is tracking employee keystrokes on Google, LinkedIn, Wikipedia as part of AI training initiative
• Lawyers should disclose when AI causes errors, appeals court says
• High earners race ahead on AI as workplace divide widens
• Companies Expand AI Health Offerings, Even as Accuracy Questions Remain — The Monitor
• Product Pages Drive SEO Discovery and AI Readiness
• UNC Charlotte introduces 'Charlie' Artificial Intelligence Chatbot to handle student tech issues
• Winning a Kaggle Competition with Generative AI–Assisted Coding
• Nigeria artist uses AI to preserve history