While AI agents flooded GitHub with a 14x traffic surge and Anthropic's Model Context Protocol crossed 97 million installs, the regulatory and legal systems scrambled to keep pace. The first US deepfake conviction landed under the Take It Down Act, OpenAI paused a UK data centre deal over energy costs, and Microsoft quietly labelled Copilot as being for "entertainment purposes only." For project delivery professionals, the signal is consistent: AI is scaling faster than the frameworks designed to manage it. The stories below are worth tracking.
Agentic AI & Developer Infrastructure
Anthropic's Model Context Protocol (MCP) Crosses 97 Million Installs: MCP is becoming the de facto standard for connecting AI agents to business systems, with every major AI provider now shipping compatible tooling.
Cursor 3: Multi-Agent IDE with Built-in Browser and Plugin Marketplace: Cursor's revamped IDE lets developers run multiple AI agents simultaneously across projects, signalling a shift from line-by-line coding to multi-agent orchestration.
Poke Makes Using AI Agents as Easy as Sending a Text: A new interface strips the complexity from agent deployment, targeting non-technical users who want AI assistance without configuration overhead.
Claude Code Source Code Leak: 512K Lines, DMCA Takedowns, and a Copyright Question: An npm packaging error exposed Claude Code's full TypeScript source, revealing its compression architecture and raising questions about copyright protection for AI-generated code.
Anthropic Says Claude Code Subscribers Will Need to Pay Extra for OpenClaw Usage: Anthropic introduced additional charges for OpenClaw support in Claude Code, signalling how agent-era pricing models are evolving.
OpenAI Adds New $100/Month ChatGPT Subscription Tier for Heavier Codex Use: OpenAI's new pricing tier targets developers and power users who need higher capacity for AI coding workloads.
Marc Andreessen: AI Agents Will Have Bank Accounts: Andreessen outlined the architecture behind agent-based AI on Latent Space, noting that the x402 Foundation launched with Coinbase, Stripe, Visa, Google, and Microsoft to embed payment rails for agent transactions.
AI Startup Rocket Offers Vibe McKinsey-Style Reports at a Fraction of the Cost: An Indian startup is using AI to generate consulting-grade analysis at a price point that undercuts traditional firms, testing how far AI can compress knowledge work.
AI in the Built Environment
DPR Construction Scales AI Procurement Platform Across 100+ Projects: One of the largest real-world deployments of AI in construction procurement, automating supplier outreach, bid comparison, and award workflows at scale.
"AI Is No Longer a Tool. It Is Becoming the Operating System of Project Delivery": PM Global argues that AI is evolving into the foundational layer through which all project management functions are accessed, automated, and executed.
AI in Contract Management Whitepaper: Built Intelligence's whitepaper examines how AI is being applied to contract administration and management workflows in construction.
Safety, Governance & Legal
NYT Opinion: Claude Mythos Could Let Hackers Break Every Major Software System: Thomas Friedman's column captures the mainstream anxiety around Mythos, framing Anthropic's decision to restrict public access as a significant moment for AI governance.
First US Deepfake Conviction Under the Take It Down Act: An Ohio man became the first person convicted under the federal law criminalising AI-generated intimate imagery, establishing legal precedent for generative AI misuse.
OpenAI Releases a New Safety Blueprint to Address the Rise in Child Sexual Exploitation: OpenAI published detailed safety protocols for tackling AI-facilitated CSAM, responding to growing regulatory and public pressure.
Yes, Microsoft Really Said Copilot Is "for Entertainment Purposes Only": Microsoft's disclaimer reveals a tension at the core of enterprise AI adoption: the gap between marketing promises and legal liability.
Infrastructure, Compute & Energy
AI Companies Are Building Huge Natural Gas Plants to Power Data Centres: Microsoft, Meta, and Google are investing in dedicated natural gas power generation for AI data centres, raising questions about sustainability commitments.
OpenAI Pauses UK Data Centre Deal Over Energy Costs and Regulation: The UK's regulatory environment and energy pricing are creating friction for hyperscale AI infrastructure investment, with potential knock-on effects for UK construction demand.
Google Cloud: 95% of AI Pilots Fail: Yasmeen Ahmad of Google Cloud argues the 95% failure rate is primarily a data quality and accessibility problem, with 90% of enterprise data remaining unstructured and invisible to AI systems.
Meta Debuts the Muse Spark Model in a Ground-Up Overhaul of Its AI: Meta's new model represents a fundamental rebuild of its AI stack, with implications for how the company competes in the foundation model landscape.
AI, Economy & Society
OpenAI's Vision for the AI Economy: Public Wealth Funds, Robot Taxes, and a Four-Day Workweek: Sam Altman argues that AI and robotics productivity could fund universal basic income and enable a shorter working week, reigniting the debate on how automation's economic gains get redistributed.
Gen Z AI Use Stable But Resentment Growing, Gallup Finds: 51% of Gen Z now use AI regularly, but excitement has collapsed from 36% to 22%, while anger toward AI has risen 9 points to 31%. Worth tracking for anyone managing workforce adoption.
How China Fell for a Lobster: What an AI Assistant Tells Us About Beijing's Ambition: The BBC examines what a consumer AI assistant reveals about China's broader AI strategy and the cultural dynamics of adoption in a tightly controlled information environment.
AGIBOT Ships 10,000th Humanoid Robot: 5,000 in Just 90 Days: Chinese robotics company AGIBOT's manufacturing velocity illustrates the pace at which humanoid robots are moving from pilot projects to large-scale commercial deployment.
Waymo Hits 500,000 Paid Rides Per Week: Waymo's weekly ride volume has increased more than tenfold in a year, representing one of the clearest examples of autonomous AI scaling to mainstream consumer use.
That’s all we have this week. More such updates coming your way in the next edition.
All content reflects our personal views and is not intended as professional advice or to represent any organisation.
