AI Pulse · Updated every Monday

The weekly AI briefing for working professionals.

Five minutes. Zero hype. Every Monday I hand you the AI news that actually matters to your career — and the one thing to do about it this week.

Skip to this week →

This week Week of Aug 4–10, 2026

What actually happened in AI this week.

Curated from the noise. Every story ends with why it matters for your career specifically.

★ Headline story

AI Education Market Hits $7.57B — Set to Exceed $112B by 2034 as Student Usage Reaches 92%

Global student AI usage jumped from 66% in 2024 to 92% in 2025 — the fastest adoption curve ever recorded for an educational tool. The global AI-in-education market hit $7.57 billion in 2025 and is forecast to exceed $112 billion by 2034. A D2L and Tyton Partners survey found more than half of higher education administrators, instructors, and students now use AI weekly — yet only 32% of institutions have a central AI policy. Faculty citing student cheating as a top challenge nearly doubled: from 36% to 55% since 2024.

Why this matters to you: If you work in corporate L&D, HR, or any training-adjacent role, you are operating in the fastest-growing sector of enterprise software. The 68% of institutions without a formal AI policy are writing that policy right now. The professionals who already know what should be in it — acceptable use, disclosure rules, tool governance — are in high demand. That’s a positioning opportunity available this week.

Regulation

EU AI Act Extends Reach to Classrooms — Student Assessment Tools Now Classified High-Risk

Following the August 2 enforcement deadline, the EU AI Act’s high-risk classification now explicitly covers AI used for student assessment, admissions screening, and academic progress monitoring. Any platform making automated decisions about learners must meet strict transparency and human oversight requirements immediately. The State University of New York simultaneously adopted a formal systemwide AI policy for all 64 campuses, requiring AI literacy training for all incoming undergraduates starting Fall 2026.

Why this matters to you: Every HR performance tool, employee assessment platform, and corporate training LMS that uses AI scoring is now in the same regulatory category as student admissions systems. If your company operates in the EU or serves EU clients, your AI-powered evaluation tools need documentation. Ask your HR tech vendor what data the model uses and whether it qualifies as high-risk — before your legal team asks you.

Research

The OECD “Illusion of Learning” Warning — AI Over-Reliance Is Now a Measurable Problem

The OECD issued a formal caution against what it calls the “illusion of learning”: students and professionals who delegate thinking to AI show measurable declines in critical thinking and metacognitive skills. Students who submit AI-drafted work receive grades without building knowledge. The same dynamic is appearing in corporate settings. Separately, a 2025 Center for Democracy and Technology report confirmed 85% of teachers and 86% of students used AI in the preceding school year — the first time student usage outpaced teacher usage.

Why this matters to you: The OECD finding applies directly to professional work. If you let AI write your reports, draft your analysis, and frame your recommendations without genuinely engaging with the output, you erode the domain expertise that makes you hard to replace. The competitive advantage in 2026 isn’t using AI — it’s using AI to amplify judgment you actually have.

EdTech & L&D

Hyper-Personalized AI Tutors Go Mainstream — and Corporate Learning Is Next

In 2026, AI tutoring platforms now adapt in real time to each learner’s pace and understanding, using Socratic questioning to prompt reasoning rather than deliver answers. Multimodal AI interfaces — text, voice, image, and video in a single session — are now standard on leading platforms. EdVisorly secured a $13.3 million Series A (total $22M) for its AI admissions automation platform, Eddie AI, signaling that venture capital is following the personalization trend from K–12 into higher education and corporate onboarding.

Why this matters to you: The AI tutoring model that’s reshaping classrooms is 12–18 months from landing in your company’s onboarding and upskilling programs. Professionals who pilot these tools internally — and document what works — become the go-to resource when the formal rollout happens. That is a very low-cost way to own a visible position on your team right now.

Enterprise

Only 32% of Organizations Have a Formal AI Policy — The Other 68% Are Creating One Right Now

Across higher education and the enterprise, the same pattern is emerging: AI adoption is running far ahead of governance. A D2L/Tyton survey found 32% of institutions have a central AI policy — meaning 68% are operating without one. IBM’s AI Operating Model blueprint, released this quarter, identifies AI policy as the single biggest predictor of successful AI adoption. Companies without it face inconsistent use, IP exposure, and compliance gaps as EU AI Act enforcement scales.

Why this matters to you: If your organization is in the 68%, someone is about to write that policy. That person will have meaningful visibility with leadership. If you understand the landscape — what tools are in use, where the risks are, what other organizations have done — you can be that person. MindBot Academy’s weekly briefings give you exactly the background you need.

Jobs market

AI Roles in Governance and L&D Are Growing as Traditional Roles Contract

July data showed AI-attributed layoffs continue across finance, research, and customer service — the roles most exposed to AI output replication. But the same restructuring announcements contain a consistent pattern: new roles in AI governance, AI operations management, learning experience design for AI-native teams, and digital transformation strategy. All require domain expertise combined with AI fluency — not engineering skills. McKinsey’s own July restructuring included the creation of an AI practice lead role in three practice areas.

Why this matters to you: The professionals at risk produce outputs AI replicates cheaply. The professionals in demand judge, govern, and direct AI output using expertise that cannot be easily automated. The gap between those two profiles is smaller than most people think — and MindBot Academy exists to help you cross it deliberately.

🔧 Tool pick of the week

NotebookLM — Turn Any Document Into an Interactive Research Assistant

Google’s NotebookLM lets you upload PDFs, reports, slide decks, or web articles and then chat with them as a grounded research assistant — it only answers from your uploaded sources and tells you exactly where each answer comes from. The Audio Overview feature converts any set of documents into a realistic two-person podcast discussion you can listen to during a commute. No hallucinations from outside your sources. How to try it: go to notebooklm.google.com (free), upload any document you’ve been meaning to read, and ask it to summarize the three most important points for your job.

Why this matters to you: If your role involves reading lengthy reports, policy documents, or research — and most professional roles do — NotebookLM compresses a 60-page PDF into a 10-minute briefing you can trust. It is the fastest way to arrive at a meeting already knowing the document better than most people in the room.

💡 Career tip of the week

Volunteer to Draft Your Team’s AI Policy — Even a One-Page Version

Only 32% of organizations have a formal AI policy. If yours doesn’t, there’s a draft waiting to be written — and the person who writes it gets visibility, credibility, and a concrete AI-leadership credential in one move. The one-page version covers: which AI tools are approved, what data can be shared with them, how AI-generated content should be disclosed, and who to call with questions. You don’t need to be a lawyer or an engineer. You need to know the landscape — which you do, if you’re reading this.

Why this matters to you: When the OECD, IBM, and EU regulators all name AI governance as the critical missing layer in enterprise AI adoption, being the person who fills that gap at your organization is not a minor contribution. It is the kind of visible, domain-relevant AI work that shows up on performance reviews, in promotion conversations, and on your LinkedIn profile.

Get next Monday's Pulse in your inbox.

One email a week. The free AI toolkit on signup. Unsubscribe any time.

Past briefings.

Every Monday since launch. The archive compounds — come back weekly to see how the story evolves.

Earlier editions

Week of July 28, 2026

EU AI Act Enforcement Begins — Frontier Model Grace Period Ends Aug 2

The EU AI Act’s grace period for general-purpose AI model makers officially ended, with the European Commission gaining power to fine developers of the most powerful models. Claude Sonnet 5 pricing confirmed rising September 1. AI-attributed layoffs hit 165,000 in 2026, with Visa cutting 2,600 roles explicitly citing AI. Microsoft Copilot added Agent Mode to PowerPoint and full-inbox reasoning in Outlook.

What we covered: EU AI Act enforcement live · Claude Sonnet 5 price rise Sept 1 · July layoff spike (48% vs June) · Visa/Intel/Samsung restructurings · Microsoft Copilot July update · Chinese model enterprise risk · AI governance role emergence

Week of July 27, 2026

Kimi K3 Open Weights Drop — The Largest Freely Available AI Model Ever

Moonshot AI released Kimi K3’s full weights (2.8T parameters, free to self-host under MIT license) while Monday.com cut 630 jobs in an explicit AI-agent pivot and Thomson Reuters announced 500 engineering eliminations tied to AI research automation.

What we covered: Kimi K3 open weights · Monday.com AI layoffs · Thomson Reuters engineering cuts · MCP agents in M365 apps · AWS $1B embedded AI deployment

Week of July 20, 2026

China’s AI Agent Law Takes Effect — The World’s First Legal Framework for Autonomous AI

China’s Implementation Opinions on intelligent agents became enforceable July 15, establishing the first dedicated regulatory category for AI agents globally. AI job cuts hit a record fourth consecutive month — but Gartner data showed no ROI advantage for companies making the deepest cuts.

What we covered: China AI Agent Law · Kimi K3 MoE model launch · Record AI-attributed layoffs · ChatGPT Work for non-technical pros · MCP standard adoption · FTC AI accuracy policy

Week of July 14, 2026

GPT-5.6 Triple Launch Triggers AI Price War — The Biggest Model Week of 2026

OpenAI, SpaceX AI, and Meta launched three flagship models within 24 hours, crashing inference costs to historic lows and triggering an industry-wide price war. Claude Cowork and ChatGPT Work also went live, bringing persistent background task delegation to non-technical professionals.

What we covered: GPT-5.6 Sol/Terra/Luna · Grok 4.5 · Muse Spark 1.1 · Claude Cowork launch · 302 AI-attributed layoff events · EU AI Act simplification · AI video production costs down 60%

Week of June 16, 2026

GPT-5.4 Launches with Computer-Use — and Claude Sonnet 4.6 Becomes the Default for Millions

OpenAI’s GPT-5.4 brought native computer-use capabilities to market — AI that can click, type, and navigate apps on your behalf. Anthropic simultaneously made Claude Sonnet 4.6 the default on claude.ai, extending improved instruction-following to free users.

What we covered: GPT-5.4 computer-use launch, Claude Sonnet 4.6 rollout, MiniMax M3 price collapse, 267 layoff events, Perplexity Deep Research going free, bipartisan US AI bill draft.

Week of May 23, 2026

Google I/O 2026: Gemini Is Now Inescapable in Every Google Product

Google announced Gemini 3.5 Omni at I/O 2026 — a single multimodal model powering Search AI Mode (1B monthly users), Gmail Daily Brief, Docs, Sheets, and the new Antigravity 2.0 agent desktop app. The “optional AI” era for Google Workspace users ended this week.

What we covered: Gemini 3.5 Omni, GPT-5.5 agentic capabilities, 142K tech layoffs, Colorado AI Act, Big Four enterprise AI deals.

Week of May 5, 2026

Anthropic, Goldman Sachs and Blackstone Launch a $1.5B Enterprise AI Firm

Anthropic partnered with Goldman Sachs, Blackstone, Apollo, and General Atlantic to launch a $1.5 billion firm that embeds AI engineers inside companies to redesign workflows and build custom agents. OpenAI announced a near-identical structure with TPG and Bain Capital.

What we covered: AI consulting arms vs. Big 4, May layoff wave (128K+), EU AI Act August deadline, Perplexity Personal Computer for Mac, IBM AI Operating Model blueprint.

Week of April 28, 2026

Big Tech’s $725B AI Splurge Is Being Funded by Mass Layoffs

Amazon, Meta, and Microsoft collectively cut over 50,000 jobs in April — with 48% of cuts explicitly attributed to AI-driven workflow automation — while Big Tech’s combined AI capex hit $725 billion for the year.

What we covered: AI-attributed Big Tech layoffs, GPT-5.5 agentic capabilities, Senate AI job-reporting bill, enterprise agent adoption (79% of orgs deployed), MS365 Copilot Agents launch.

Week of April 21, 2026

Meta + Microsoft cut 20,000+ jobs in 24 hours — both explicitly cite AI automation

Meta announced 8,000 layoffs and Microsoft offered buyouts to ~12,000 US employees in the same 24-hour window. Both companies pointed directly to AI automation in content moderation, software testing, and customer support as enabling the cuts.

What we covered: AI-attributed layoffs, DeepSeek V4 open-source launch, AI-related legislation, PwC AI value concentration study, Google enterprise agent toolkit.

Week of April 21, 2026

Claude Opus 4.7 launches — Anthropic reclaims the top LLM spot

Anthropic released Claude Opus 4.7 on April 16 with agentic task budgets, a /ultrareview command, and Claude Design for visual one-pagers. It scored 64.3% on SWE-bench Pro and topped the GDPval-AA Elo benchmark.

What we covered: Opus 4.7 launch, GPT-5.4 vs Gemini 3.1 Pro benchmarks, Oracle 30K layoffs, Stanford AI Index, White House AI Policy Framework, Adobe Firefly.

Week of April 13, 2026

Anthropic’s revenue passes OpenAI — $30B vs $25B annualized

Anthropic announced its revenue run rate crossed $30B, overtaking OpenAI for the first time, with over 1,000 enterprise customers spending >$1M/year on Claude. Both Claude Opus 4.7 and Gemini 3.1 Ultra were announced this week.

What we covered: Enterprise AI spending race, Opus 4.7 & Gemini 3.1 Ultra launches, Q1 layoff data, Google Sheets AI productivity gains, Adobe Firefly launch.

Week of April 6, 2026

OpenAI, Anthropic, Google unite against model copying

The three US AI leaders announced a coordinated intelligence-sharing effort via the Frontier Model Forum to stop Chinese firms (DeepSeek, Moonshot, MiniMax) from adversarial distillation of their models.

What we covered: Geopolitics of AI, MCP crossing 97M installs, Anthropic’s Coefficient Bio acquisition, life sciences expansion.

Week of March 30, 2026

GPT-5.4 Thinking + Gemini 3.1 Pro benchmark sweep

OpenAI’s GPT-5.4 took the versatility crown; Claude Opus 4.6 and Gemini 3.1 Pro both topped 50% on Humanity’s Last Exam. Frontier model progress visibly accelerated.

What we covered: What "reasoning models" actually do in practice, benchmark literacy, and when to pay for Pro vs use free tiers.

More editions coming as the archive grows. Subscribers get every new one delivered.