AI @ Work
Friday, May 15, 2026
Issue #006
---
This week: seven in ten Americans don’t want AI data centers near them — and a Nobel laureate may have part of the solution. Claude launches fifteen ready-to-run workflows for small businesses. Two out of three US doctors are now using an AI tool most people have never heard of. And Princeton abandons a 133-year-old honor code because of AI. All translated into what it actually means for your work.
---
★ This Week’s Big Story
Why this matters to you
7 in 10 Americans oppose AI data centers in their communities — and only 16% use AI regularly. That gap is the industry’s real problem.
A new Gallup poll found that 70% of Americans oppose AI data center construction near their homes — bipartisan, intense, and growing. The opposition cuts across political lines and age groups. Water consumption is the most cited concern: data centers are on track to guzzle 68 billion gallons of water annually by 2028, a 300% increase in five years. Only 16% of Americans use AI tools regularly, which means the industry is asking communities to bear enormous costs — noise, heat, water, power draw — for a technology most of them have never personally used.
There may be a partial solution from an unlikely direction. Atoco, a startup founded by Nobel Prize-winning chemist Omar Yaghi, has built a shipping-container-sized machine that pulls 4,000 liters of water directly from the atmosphere per day — even in arid regions — using microscopic structures that trap water molecules at the atomic level. It can run on waste heat from data center machinery, turning a cooling problem into a water source. Production starts later this year. What the Gallup poll tells you: AI adoption has a public trust deficit, and the industry hasn’t figured out how to close it. | Executive & Strategy | Legal & Compliance | All Readers |
Source: Superhuman Newsletter, May 14, 2026 · Gallup · Atoco
---
AI Platforms
Claude / Anthropic
---
If you run a small business or a law firm, Anthropic just built fifteen workflows specifically for you — connected to the tools you already use.
Claude for Small Business Launches: QuickBooks, PayPal, HubSpot, Canva, DocuSign — All in One Conversation
Anthropic launched Claude for Small Business this week, plugging Claude Cowork directly into QuickBooks, PayPal, HubSpot, Canva, DocuSign, Google Workspace, and Microsoft 365. Fifteen ready-to-run agentic workflows come out of the box — from planning payroll and chasing invoices to launching marketing campaigns. Anthropic also rolled out twelve one-click workflows specifically for legal professionals. Daniela Amodei, Anthropic’s co-founder, framed it plainly: small businesses are 44% of US GDP but have lagged far behind large enterprises in AI adoption. This is Anthropic’s attempt to close that gap. A free PayPal course for small business AI adoption is included with the launch. | Operations | Finance | Legal & Compliance |
Source: Superhuman Newsletter, May 14, 2026
---
ChatGPT / OpenAI
---
If you use Codex to run AI tasks, you can now monitor and control them from your phone while doing something else entirely.
Codex Goes Mobile — Monitor Your AI Agents, Approve Commands, and See Screenshots From Your iPhone
OpenAI released a mobile interface for Codex this week, letting you manage running AI agents directly from your iPhone or Android. You can see real-time screenshots of what your agent is doing, approve or reject commands before they execute, and redirect tasks mid-run — all from your phone. You kick off a long research or coding task on your desktop, step into a meeting, and stay in control from your pocket. This is a meaningful shift in how agentic AI actually fits into a workday — it no longer requires you to be at your desk watching it run. | Operations | IT & Security | Executive & Strategy |
Source: The Deep View, May 13, 2026
---
For anyone tracking AI governance — OpenAI just made a surprising proposal to the US government this week.
OpenAI Proposes a Global AI Watchdog Modeled on the Nuclear Regulatory Agency — With China Inside It
OpenAI told US lawmakers this week that it supports creating an International Atomic Energy Agency-style oversight body for AI — with the United States leading it and China as a participating member. This is a significant shift from the America-first framing that has dominated US AI policy. OpenAI’s position: the risks of powerful AI are global, and excluding China from governance frameworks creates more danger than it prevents. Whether Congress agrees is another matter, but it’s the clearest signal yet that the leading AI company sees international cooperation — not competition — as the long-term answer to AI safety. | Legal & Compliance | Executive & Strategy |
Source: Latestly AI, May 14, 2026 · Bloomberg
---
Cross-Platform
---
There’s an AI tool that two thirds of US doctors use every day — and most people have never heard of it.
OpenEvidence Is Now Used by 650,000 US Physicians. It Does One Thing. That’s the Point.
A company called OpenEvidence has quietly become infrastructure for American medicine. It does one narrow thing: answers medical questions for doctors — drug interactions, treatment protocols, dosage recommendations — drawing on peer-reviewed research in real time. As of this week, roughly 650,000 practicing physicians in the United States use it, about two thirds of all US doctors, plus 1.2 million internationally. The lesson for every organization thinking about AI adoption: the tools that last aren’t the ones that impress people — they’re the ones that become impossible to start your morning without. | HR & People | Executive & Strategy | Operations |
Source: Latestly AI, May 14, 2026
---
Tools & Workflow — Try These This Week
---
For anyone who uses Claude Cowork and wants to get up and running in three steps.
The Claude Cowork Quick-Start: Three Steps to Organize Your Files, Summarize Your Documents, and Draft Your Emails
Step one: connect your Google Drive or OneDrive and ask Claude to organize your files by project or date — it will sort and rename in seconds. Step two: upload any document — contract, report, meeting notes — and ask Claude to give you a one-paragraph summary and three action items. Step three: describe an email you need to send, including the recipient’s context and what outcome you want, and let Claude draft it. Most people who try all three in one session report that Step three is where they feel the biggest time savings — drafting emails that require political sensitivity or persuasive framing is where Claude earns its subscription fee. | Operations | HR & People | All Readers |
Source: The Deep View, May 13, 2026
---
For anyone who wants to ask AI about sensitive topics — health, money, relationships — without leaving a record anywhere.
WhatsApp Just Launched a Private AI Chat Mode — Not Even Meta Can Read It
Meta rolled out an incognito mode for AI chats inside WhatsApp this week. When switched on, your messages to Meta AI and its responses are not stored on Meta’s servers — and the chat history disappears from your side too. The feature was built specifically for people who want to ask AI about sensitive topics — health symptoms, financial situations, relationship problems — without that information being saved anywhere. Meta CEO Mark Zuckerberg called it the first major AI product with no chat logs stored on servers. It’s text only for now — but for genuinely private AI conversations, it’s a meaningful option. | Legal & Compliance | HR & People | All Readers |
Source: Mindstream, May 14, 2026 · BBC News
---
Worth Watching
Princeton University abandoned its 133-year-old honor code this week — because of AI. The university is bringing back supervised exams for the first time since 1893, after faculty voted that AI tools, laptops, and phones have made cheating too easy to catch. Nearly 30% of Princeton seniors surveyed admitted to cheating on an assignment or exam. AI didn’t create academic dishonesty — it just made the old system of trusting students on their honor untenable. | All Readers |
Cerebras went public this week at $350 per share, closing with a $95 billion market cap. Cerebras makes AI chips designed to compete with NVIDIA — faster for certain AI workloads, purpose-built for inference rather than training. A $95 billion debut for a chip company most people outside the industry have never heard of signals how much institutional money is chasing AI infrastructure right now. | Finance | Executive & Strategy |
Anthropic and the Gates Foundation pledged $200 million jointly for AI in health and education. The money will fund AI tools for under-resourced schools and healthcare systems in low-income countries — the largest single AI-for-good commitment from a major AI company and a major philanthropy working together. | Executive & Strategy | HR & People |
“Vibe coding” tools are leaking sensitive data — 380,000 files exposed. AI coding tools like Lovable and Replit, which let non-technical staff build production software without programming knowledge, are producing apps with basic security gaps. Healthcare records, financial data, and internal systems are among the exposed files. If your organization is using AI to build internal tools, a security review is not optional. | IT & Security | Legal & Compliance |
---
Published by Independent Intelligence — an independent AI newsletter for people who use AI at work, not just read about it. Five newsletters, one Substack. Friday general edition + Wednesday deep-dives on Claude, ChatGPT, Gemini, and Perplexity. Forward freely.