|
|
| FEATURED |
| ChatGPT Work Just Became Your New Coworker — And It Ships With GPT-5.6 |
|
OpenAI dropped the biggest workplace AI agent yet on Thursday. ChatGPT Work, powered by the new GPT-5.6 model family, doesn’t just answer questions — it executes multi-step projects across your connected apps, files, and desktop tools. Think: gathering data from Slack and Google Drive, building a financial model in Sheets, and turning it into a presentation — all while you’re in a meeting.
Why it matters for your business: This isn’t a chatbot upgrade. It’s an AI employee that works for hours, breaks projects into steps, and ships deliverables. The desktop app can click, type, and move files using Computer Use — meaning it can operate legacy software that has no API. Over 5 million people already use Codex weekly; 1 million of them are outside software development. The agent era just crossed from “experimental” to “on your desktop.”
The competitive landscape: Anthropic’s Claude Cowork already triggered a selloff in software stocks earlier this year. Now OpenAI is shipping to Pro, Enterprise, and Edu users immediately — with Plus and Business following within days. Microsoft Copilot Cowork and Google’s Genkit Agents are also in the arena. The workplace AI agent wars are officially underway, and the winners will be the businesses that learn to orchestrate these agents — not just deploy them.
|
|
📎 USA Today | AppleInsider
|
|
| 🇬🇧 UK Deploys AI Agents to Fight AI-Powered Cyberattacks at National Scale |
|
The UK’s NCSC launched Cyber Shield on July 7 — a national-scale plan to deploy agentic AI for autonomous cyber defense. The blueprint: AI red teams find vulnerabilities in minutes (not weeks), while AI blue teams detect and contain breaches across government and industry boundaries. Simultaneously, 60 companies signed the Cyber Resilience Pledge, committing to board-level accountability and supply chain security.
So what? The same AI agents you’re deploying to automate customer service are also being deployed to attack your infrastructure. The speed gap is the story: AI-assisted attacks now find vulnerabilities in minutes. If your defenses still run on human-paced patching cycles, you’re already behind. The UK is betting that only AI can defend against AI — a thesis every business with a digital footprint needs to understand.
|
|
📎 SecurityWeek
|
|
| 🔓 69% of Enterprises Share AI Agent API Keys — And They’re Underestimating Failure Rates by 2.25x |
|
Two sobering findings landed this week. First, a VentureBeat survey found that 69% of enterprises share AI agent credentials — meaning one compromised agent can inherit access to every system. Second, a study of 67 AI models found enterprises underestimate multi-model failure rates by 2.25x. In plain English: your AI agents are more likely to fail than you think, and when one does, it might have the keys to everything.
The practical fix: If you’re deploying AI agents, credential isolation isn’t optional — it’s table stakes. Each agent should have the minimum permissions it needs, and no two agents should share an API key. As for failure rates: build a kill switch, test for cascading failures, and assume your agents will fail in ways you didn’t predict. The enterprises getting ROI from AI are the ones that plan for failure — not the ones that pretend it won’t happen.
|
|
📎 VentureBeat
|
|
|
|
| Meta Ships Muse Spark 1.1 — Its First Paid AI API, Taking the Gloves Off Against OpenAI |
|
Meta launched Muse Spark 1.1 on Wednesday — its most capable model for coding and agentic tasks, and crucially, its first-ever paid API. The specs: one-million-token context window, multimodal (text, images, video), and pricing at $1.25/M input tokens — cheaper than Claude Sonnet 4.6, pricier than GPT-5 mini. Mark Zuckerberg broke a 3-year silence on X to announce it. Early partners include Replit, Cline, and Box.
The strategic shift: Meta has always been the open-source champion. Charging for API access signals a new era: the company now believes its models are competitive enough to monetize directly — not just as a loss leader for its ad business. For businesses evaluating AI providers, the landscape just got more interesting: you now have three credible frontier API providers (OpenAI, Anthropic, Meta) with meaningfully different pricing and capabilities.
|
|
📎 Hindustan Times
|
|
| 🏛️ Fed Chair Taps Marc Andreessen to Co-Lead AI Economic Impact Task Force |
|
Federal Reserve Chair Kevin Warsh announced five external task forces on Wednesday — and tapped Marc Andreessen (a16z) to co-lead the one on productivity and jobs, alongside Stanford economist Charles I. Jones and Microsoft Xbox CEO Asha Sharma. Warsh called AI “the most disruptive moment in modern economic history” and wants recommendations by year-end.
Why this matters for interest rates and your business: Warsh’s thesis is that AI could boost productivity while putting downward pressure on long-term inflation. If the task force validates this, it could reshape Fed thinking on interest rates — potentially supporting more accommodative monetary policy. For business owners, the signal is clear: the highest levels of economic governance are now treating AI not as a tech trend, but as a macroeconomic force.
|
|
📎 Crypto Briefing
|
|
| 🇨🇳 China Moves to Restrict AI Model Exports — Open-Source Models Like DeepSeek R1 in the Crosshairs |
|
Beijing is preparing to restrict exports of advanced AI models, including both closed-source and open-source models, according to reports on July 10. Discussions are underway with Alibaba, ByteDance, and Z.ai. The policy would classify AI technology leaks as national security violations. Chinese open-weight models like DeepSeek’s R1 have gained significant adoption in the U.S. market over the past year due to competitive pricing and capabilities.
The bottom line: If implemented, this fragments the global AI supply chain further. Businesses relying on Chinese open-source models may need contingency plans. The move mirrors U.S. chip export controls — but from the other direction. AI is now officially a strategic asset in a new tech Cold War, and every business operating across borders needs to pay attention.
|
|
📎 Tempo.co
|
|
|
🚀 Want AI working for YOUR business? Most companies are experimenting with AI chatbots. We deploy AI workforces — AI Employees that follow up on leads, resolve support tickets, publish content, chase invoices, and screen 200 job applicants overnight so your hiring manager starts Monday with the top 10. Each role has a cost profile and human oversight, managed through one platform. This newsletter? Written by an AI Employee, approved by a human — so our team stays focused on what only humans can do. AIToken Labs helps businesses design their AI Workforce Operating Model — starting with the 2-3 roles that deliver ROI in the first 60 days. Book a free 40-minute AI Workforce Blueprint Session →
|
|
