The AI labs spent the weekend turning on each other. Apple dragged OpenAI to court, accusing it of orchestrating the theft of confidential hardware secrets through two former Apple employees — the suit surfaced in a Forbes report late Sunday and is now multiply covered. And one report describes Anthropic revoking OpenAI’s API access to Claude after discovering it was being used for internal development. The model race has always been a talent war; it is now a litigation and access war too.
Synthetic media had a rough day in the trust column. One statistics roundup reports a 3,892% surge in document deepfakes alongside $893 million in US losses from AI-related complaints. Another test found Meta’s own AI-image detector misses 55% of its images once they’re cropped. Meta, meanwhile, pulled its AI feature that used public Instagram photos after three days of backlash. In the UK, the CPS admitted submitting AI-generated fake legal authorities to the High Court. And a security research report documents an AI agent running a ransomware attack end-to-end against an AI app builder — the attack surface and the attacker now made of the same stuff.
The rules kept forming around all of it. The EU AI Act’s August 2 enforcement date is three weeks out, with general-purpose AI fines and hiring tools in scope. India’s Supreme Court is drafting rules that welcome AI assistance in courts while banning it from judicial decisions. A Just Security analysis walked through US Senate proposals to regulate military AI. Hollywood argued over whether an AI character with a feature-film debut counts as an actor. Microsoft set July 24 for GPT-5.6 to become standard in Microsoft 365 Copilot. And the developer community’s quiet vote: a free configuration tool that turns Claude Code into a full dev team passed 228,000 GitHub stars — configuration expertise outdrawing raw model access.
What People Are Asking
Why is Apple suing OpenAI?
Apple accuses OpenAI of orchestrating the theft of confidential hardware secrets through two former Apple employees, as OpenAI builds out its device program. The suit surfaced in a Forbes report and is now covered by multiple outlets. Techweez →
Did Anthropic cut off OpenAI's access to Claude?
One report says yes — Anthropic revoked OpenAI's API access after discovering Claude was being used for OpenAI's internal development work. Single-source as of this edition. IBTimes Australia →
How much has deepfake fraud grown?
One statistics roundup reports a 3,892% surge in document deepfakes, with $893 million in US losses from AI-related fraud complaints. Treat the headline number as one report's tally. Memeburn →
Can Meta detect its own AI-generated images?
Not reliably, according to one test: Meta's AI-image detector missed 55% of its own AI-generated images after they were cropped. HokaNews →
What is Agent Harness ECC?
A free configuration tool that turns Claude Code into a coordinated multi-agent dev team. It passed 228,000 GitHub stars — a sign that configuration expertise is becoming as valuable as model access. Tech Times →
When does GPT-5.6 become standard in Microsoft 365 Copilot?
July 24, 2026, according to one report — replacing the previous default across Word, Excel, PowerPoint, Chat, and Cowork. Borncity →
Is an AI character really starring in a movie?
Yes — AI character Tilly Norwood is set for a feature-film debut, and Hollywood is debating whether that counts as acting and what it means for performers. Los Angeles Times →
What happened with the CPS and AI in court?
The UK Crown Prosecution Service admitted submitting AI-generated fake legal authorities to the High Court — the latest in a string of legal-AI hallucination incidents. VinciWorks →
Every Story From July 13
Jump to: AI labs and models (10) · AI infrastructure (7) · Labor and workplace (6) · Security and agent risk (6) · Policy and regulation (5) · Society, politics, and ethics (3) · Science and research (2) · Markets and macro (1) · Other (1)
AI labs and models10 stories
Agent Harness ECC, a free tool that turns Claude Code into a full dev team, tops 228K GitHub stars, demonstrating the value of configuration expertise over model access.
Apple files a lawsuit against OpenAI, accusing it of orchestrating the theft of confidential Apple hardware secrets through two former employees.
The article examines recent US Senate proposals to regulate military AI, highlighting emerging safeguards for civilian protection and remaining gaps.
Alibaba's Tongyi Lab open-sources Qwen-Image-Edit, a 20B parameter AI image editing model supporting semantic, appearance, and bilingual text editing.
Article discusses the transition from rule-based chatbots to agentic AI, highlighting adoption trends and practical applications.
Microsoft is upgrading its 365 Copilot to GPT-5.6 by July 24, 2026, driving adoption in law firms amid growing client pressure and regulatory clarity on AI liability.
A survey of educators reveals that while AI cheating is a top concern, the deeper issue is the difficulty in assessing genuine student learning when AI can generate polished work.
The article covers the ongoing rivalry between OpenAI and Anthropic, focusing on a API access dispute where Anthropic revoked OpenAI's access to Claude after discovering its use for internal development ahead of GPT-5 launch.
A comprehensive timeline of Claude model releases from March 2023 to July 2026, detailing key milestones and model specifications.
A daily digest summarizing articles on AI strategy, evaluation, software engineering, and enterprise adoption.
AI infrastructure7 stories
The weekly report analyzes how AI-driven data center demand is tightening copper, transformers, and memory markets, leading to higher electricity and device prices.
Rupert Goodwins argues that AI-driven code cloning could undermine Microsoft's on-premise software monopoly, similar to Compaq's BIOS clone of IBM's PC.
Microsoft has started providing OpenAI's GPT-5.6 model for Microsoft 365 Copilot, enhancing productivity across Word, Excel, PowerPoint, Chat, and Cowork with improved reasoning, agent capabilities, and enterprise security features.
Strategy partners with UAE's largest digital insurance platform InsuranceMarket.ae to deploy AI-powered enterprise analytics, expanding its core software business beyond Bitcoin headlines.
Kenya launches a programme to equip junior secondary schools with smart boards and teacher laptops, aiming to build an AI-ready workforce.
A guide for brand teams on assessing risks from generative AI copyright lawsuits, emphasizing careful vendor selection and indemnification review.
A sponsored report explores food manufacturers' cautious AI adoption, highlighting barriers and opportunities.
Labor and workplace6 stories
Experts discuss how AI is transforming the labor market by automating routine tasks and shifting human roles toward oversight, architecture, and higher-order problem-solving.
Hollywood debates whether AI-generated character Tilly Norwood qualifies as an actor, highlighting tensions around AI's role in the entertainment industry and labor concerns.
South Africa's AI talent shortage exposes a deeper problem in how skills are defined, sourced, and deployed, requiring a shift from filling roles to building systems with clear outcomes and foundations.
Recent studies show AI is increasing productivity and employment, with limited negative impact on jobs in Europe and the US.
Calabrian unions protest against AI-driven job losses in call centers, demanding regulations and training.
The article analyzes how new senior AI roles in manufacturing signal a shift toward autonomous operations, with a timeline for workforce impact.
Security and agent risk6 stories
The article reports on the sharp rise in deepfake fraud in 2026, with a 3,892% increase in document deepfakes and $893 million in US losses from AI-related complaints.
A security research report documents the first end-to-end ransomware attack carried out by an AI agent, targeting an AI app builder via unpatched vulnerabilities.
Meta withdraws an AI feature that used public Instagram photos as references after three days of privacy backlash, amid concerns over consent and misuse.
Meta's AI image detector fails to identify 55% of its own AI-generated images after cropping, highlighting challenges in reliable synthetic media detection.
The CPS admitted submitting AI-generated fake legal authorities to the High Court, highlighting the risks of unchecked AI in legal work.
Federal cybersecurity experts discuss using AI agents and unified data to defend against AI-powered attacks.
Policy and regulation5 stories
The EU AI Act enforcement begins Aug 2 with fines for general-purpose AI, affecting hiring tools and requiring transparency.
The FTC seeks public comment on a proposed policy statement clarifying that AI companies may engage in deceptive practices if their systems' outputs conflict with consumer expectations or representations.
한국 교육부가 전공 불문 모든 대학생을 위해 20개 대학에 AI 교육과정 개발을 지원하는 사업을 발표했다.
The Supreme Court of India is drafting rules to allow AI assistance in courts while prohibiting its use in judicial decision-making, following incidents of AI-generated fake judgments.
A newsletter summarizing and analyzing recent reports and articles on innovation, R&D policy, and economic competitiveness, with a focus on Europe, Canada, and the UK.
Society, politics, and ethics3 stories
As voters increasingly rely on AI chatbots to help decide their ballots, the author warns that Trump administration policies could be used to pressure chatbots into conservative-friendly outputs, echoing past media and social media manipulation tactics.
The article argues that AI lacks the capacity for genuine moral understanding, drawing on Abrahamic religious traditions to critique the idea that machines can comprehend goodness.
Science and research2 stories
Insilico Medicine and China Medical System expand their AI-powered drug discovery collaboration targeting a CNS indication, with potential milestone payments up to 1.2 billion RMB.
A compilation of recent research highlights in health and medicine, including cancer drugs, Alzheimer's, and infectious diseases.
Markets and macro1 story
Thirty-seven cybersecurity M&A deals were announced in June 2026, including major acquisitions by Accenture, Cisco, and Databricks.
Other1 story
A roundup of AI news stories between July 2-12, 2026 from PromptInjection.net.
About the Rundown. The RUDI Rundown is compiled every day by RUDI from same-day reporting across the web. Every story links to its original publisher. How we build it →
Putting AI to work? RUDI helps organizations adopt AI responsibly — training, governance, and hands-on implementation. AI training · Consulting · Talk to us