The EU AI Act’s August 2 date arrived, and it landed on the people who do the paperwork. Italian schools now carry binding transparency and risk-management duties — labeling AI interactions, banning emotion recognition, and their teachers have to be trained to an AI-literacy standard, because the school itself is the deployer. Dutch accountants inherited inventory, documentation, and control obligations for the high-risk systems they run, Big Four tooling included. What did not arrive is the tier everyone spent the year preparing for: the high-risk obligations were pushed back and the training-data requirements relaxed, which a German public-radio report treats as an open question about whether the Act still bites. A compliance explainer sorts the two piles — what applies now, what the Digital Omnibus postponed, and how a deployer quietly becomes a provider.
Employers got the sharper version of the same day. Poland’s recruiters woke up owing human oversight, candidate transparency, and heavy fines for skipping either — and eleven governments chose that week to warn that the candidate may be the synthetic one. Per one report, North Korean IT operatives are using real-time deepfake video to get through hiring screens, with $800 million said to be flowing back to Pyongyang’s weapons programs. A single-source figure worth reading skeptically, but the mechanism is the point: the new rules govern the machine screening the candidate, not the candidate screening as a human. Inside the companies already hiring, a Forbes contributor reads survey data showing two thirds of employees stay online to hide the time AI saves them, and argues that paying for presence instead of results is what converts a productivity gain into busywork. A European adoption guide pitches the balance employers are now supposed to strike, department by department.
The capacity argument moved east. One report has DeepSeek planning a gigawatt-scale data center in Inner Mongolia — the lab that built its reputation on training efficiency now buying physical plant. A CNBC op-ed argues the U.S. lead in AI is all but gone, and that China’s coordinated ecosystem strategy is why, while an analysis of NATO’s posture makes the supply-chain version of that case: alliance systems that cannot be inspected, running partly on Chinese components. At home the appetite for rules is not the constraint — an opinion piece citing University of Maryland surveys puts support for federal AI oversight at 79 percent among Republicans. Two numbers frame the rest: automated requests overtook human web traffic for the first time, by one measure, and a trading-desk preview expects Nvidia to post 96 percent revenue growth on August 26, which will settle more of this argument than any op-ed.
What People Are Asking
What does the EU AI Act require starting August 2, 2026?
Transparency and deployer obligations take effect — labeling AI interactions and AI-generated content, AI literacy for staff, and human oversight where AI is used on people. A compliance explainer also flags how a deployer can become a provider, which carries heavier duties. GamingTechLaw →
Was the EU AI Act delayed?
Partly. The high-risk obligations were postponed and training-data requirements relaxed under the Digital Omnibus, which one German public-radio report frames as a real question about how much the Act still constrains. Deutschlandfunk →
Do schools have to follow the EU AI Act?
In Italy, yes, as of August 2: schools must label AI interactions, manage risk, and stop using emotion recognition, and teaching staff must be brought to an AI-literacy standard because the school is the deployer. News Istruzione →
What are the new EU rules for AI in hiring?
From August 2, employers using AI in recruitment owe human oversight of the decision, transparency to candidates, and face substantial fines for non-compliance. Spider's Web →
Are North Korean workers using deepfakes to get hired?
Eleven nations issued a coordinated warning that North Korean IT operatives are using real-time deepfake video to beat hiring checks, with a reported $800 million funneled to Pyongyang's weapons programs. The dollar figure comes from one report and is worth treating as an estimate. Tech Times →
Does the EU AI Act apply to accountants?
Yes — Dutch accountants now face inventory, documentation, and control obligations for high-risk AI systems they deploy, including ChatGPT, Microsoft, and Big Four tooling, with fines for gaps. AI Insider →
Is DeepSeek building its own data center?
One report says DeepSeek is planning a gigawatt-scale data center in Inner Mongolia — a shift toward owning physical infrastructure from a lab known for training efficiency. Memeburn →
Has the US lost its AI lead over China?
A CNBC op-ed argues the lead is all but gone and attributes it to China's coordinated ecosystem strategy, calling for a comparable American approach. That is an argument, not a measurement. CNBC →
Are bots now most of web traffic?
By one measure, automated requests overtook human web traffic for the first time — a shift that puts advertising metrics and supply quality in question. PPC Land →
Every Story From August 2
Jump to: Policy and regulation (9) · AI infrastructure (6) · AI labs and models (4) · Consumer tech and culture (3) · Labor and workplace (3) · Markets and macro (1) · Other (1) · Science and research (1) · Security and agent risk (1) · Society, politics, and ethics (1)
Policy and regulation9 stories
Op-ed argues that the U.S. is losing its AI lead to China's coordinated ecosystem strategy, calling for a comparable American approach.
The EU delays high-risk AI obligations and relaxes training requirements, raising concerns about the effectiveness of the AI Act.
Dutch accountants face new EU AI Act obligations for high-risk systems, requiring inventory, documentation, and controls to avoid fines.
An opinion piece citing bipartisan majorities of Americans supporting federal and international AI regulation, based on University of Maryland surveys.
Italian schools now face binding transparency and risk-management obligations under the EU AI Act, including labeling AI interactions and banning emotion recognition, creating immediate compliance duties.
Explains what parts of the EU AI Act take effect on 2 August 2026, what the Digital Omnibus postponed, and how deployers can become providers.
This article explains that schools, as deployers under the EU AI Act, must ensure AI literacy for staff by August 2, and outlines available training paths.
An analysis arguing that Anthropic's $1.5 billion settlement for pirated books is a minor financial blow and does not change the legal landscape for AI training on copyrighted works.
A weekly AI and geopolitics news analysis podcast summarizing top trends and developments.
AI infrastructure6 stories
DeepSeek plans a gigawatt-scale AI data center in Inner Mongolia, signaling a major shift toward building physical infrastructure.
Automated requests overtook human web traffic for the first time, demanding a rethink of advertising metrics and supply quality.
A detailed pricing breakdown of Lovable and its alternatives, analyzing costs, features, and pros and cons for potential users.
This analysis examines how NATO's reliance on opaque AI systems and foreign supply chains, particularly Chinese components, creates critical vulnerabilities in its defense infrastructure.
A comprehensive reference guide comparing all Claude models from Claude 3 to Opus 5, including specs, pricing, and recommendations for developers.
A daily digest article aggregating and summarizing recent AI news and analysis, covering topics like AI price wars, security, and infrastructure.
AI labs and models4 stories
A daily AI news roundup covering various stories including AI server exports, AI citation visibility, AI ethics, AI debt markets, Chinese VC fundraising, Apple's bug report limits, OpenAI hiring a mathematician, parenting AI backlash, missed policy deadlines, Google app changes, and xAI video generation.
A live index ranking top AI models by performance, with pricing and evidence verification, from BenchLM.
This page provides a detailed breakdown of Qwen API pricing on Alibaba Cloud, including rates for different models and comparisons with competitors.
An overview of ten anticipated AI transformations by 2026, focusing on automation, ethics, and societal integration, with implications for industry and governance.
Consumer tech and culture3 stories
A detailed guide to Anthropic's April Fools 2026 Claude Buddy terminal pet, covering its 18 species, rarity tiers, stats, hats, and anti-cheat mechanics.
Uptodown provides a download page and overview for the Grok AI Android app, highlighting its uncensored responses, internet search, and image generation features.
Release notes for OpenClaw version 2026.7.2 detailing new features, changes, and fixes.
Labor and workplace3 stories
Starting August 2, the EU AI Act introduces strict regulations on AI use in recruitment, requiring human oversight, transparency, and heavy fines for violations, impacting employers' HR processes.
Forbes contributor Nirit Cohen analyzes survey data showing 66% of employees fake productivity by staying online, and argues that rewarding presence over results undermines AI adoption and capacity-building.
A comprehensive guide for European companies on practical AI adoption across departments, emphasizing the balance between efficiency and compliance with the EU AI Act.
Markets and macro1 story
Nvidia's upcoming earnings could catalyze a stock turnaround, with analysts expecting 96% revenue growth and potential upside from stronger-than-expected guidance.
Other1 story
BenchLM's updated LLM API pricing comparison and calculator for August 2026 lists prices for 129 models across 22 providers, with tools to compute costs and evaluate value.
Science and research1 story
A collection of brief summaries of recent health and medicine research findings.
Security and agent risk1 story
Eleven nations issue a coordinated warning that North Korean IT operatives are using real-time AI deepfake video to deceive hiring managers, funneling $800 million to Pyongyang's weapons programs and posing a significant security risk.
Society, politics, and ethics1 story
A personal essay reflecting on the need to soften anti-AI stances to accommodate friends who use AI as an accessibility tool, while still criticizing tech companies' push for AI ubiquity.
About the RUDI Daily. Responsible Use of Digital Intelligence, daily. Compiled each day 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