In partnership with
🤖 AI, GOVERNMENT, AND THE AGENT ERA COLLIDE
This week was messy in a very real-world way.
One lab got pushed out of federal systems over safety limits. Another stepped in and signed a classified deployment. Users responded with a boycott. Meanwhile, agent products kept creeping into everyday work.
Here are the five stories worth knowing.
⚡ Quick Overview
Anthropic gets kicked out: The government orders a phase-out after a safeguards standoff.
OpenAI signs a classified deal: Three red lines, plus a safety stack OpenAI says it controls.
QuitGPT gains steam: Users cancel subscriptions over politics, ethics, and military ties.
Notion launches Custom Agents: Triggers, schedules, cross-tool actions, free to try until May.
Perplexity launches Computer: A managed, multi-agent system that orchestrates other models.
Your AI tools are only as good as your prompts.
Most people type short, lazy prompts because writing detailed ones takes forever. The result? Generic outputs.
Wispr Flow lets you speak your prompts instead of typing them. Talk through your thinking naturally - include context, constraints, examples - and Flow gives you clean text ready to paste. No filler words. No cleanup.
Works inside ChatGPT, Claude, Cursor, Windsurf, and every other AI tool you use. System-level integration means zero setup.
Millions of users worldwide. Teams at OpenAI, Vercel, and Clay use Flow daily. Now available on Mac, Windows, iPhone, and Android - free and unlimited on Android during launch.
ANTHROPIC REFUSES THE PENTAGON’S TERMS, THEN GETS BANNED
Source: Anthropic
What’s Happening
Reporting says Anthropic rejected a Pentagon demand to loosen specific safeguards, including limits related to mass domestic surveillance and fully autonomous weapons.
The administration then ordered federal agencies to stop using Anthropic tech, with a phase-out period, and the Pentagon labeled the company a “supply chain risk.”
Anthropic says it will challenge the designation.
Why It Matters
Safety terms are now a government procurement fight. This is no longer just “model policy,” it is access to public-sector contracts.
The label has ripple effects. “Supply chain risk” can spill into contractors and partners, not just direct federal deals.
OPENAI SIGNS A CLASSIFIED DEPLOYMENT DEAL, WITH “RED LINES”
Source: OpenAI
What’s Happening
OpenAI announced an agreement to deploy advanced AI systems in classified environments.
In its own write-up, OpenAI says the deal includes three “red lines” such as no domestic mass surveillance and no autonomous weapons without human involvement, and language it says strengthens privacy protections.
OpenAI also says it retains control over key parts of its safety stack and expects a working group with other labs and partners.
Why It Matters
This sets a template other labs will be pressured to follow. If one lab can sign, others will be asked why they won’t.
People will judge the “guardrails” by what happens next. The first misuse story will travel faster than the fine print.
QUITGPT TRENDS, AND SUBSCRIPTIONS BECOME A PROTEST TOOL
Source: MIT Technology Review
What’s Happening
A “QuitGPT” campaign has been urging users to cancel ChatGPT subscriptions, fueled by ethical and political objections to AI companies’ ties to government and broader controversy around how these tools get used. Reporting also notes user frustration with quality and tone changes in newer models.
The movement has spread across social platforms and tech coverage, turning consumer behavior into a visible signal.
Why It Matters
AI is becoming values-coded. People are picking assistants the way they pick brands.
OpenAI’s business model is now part of the story. Government deals, enterprise contracts, ads, all of it gets pulled into public perception.
NOTION LAUNCHES CUSTOM AGENTS THAT RUN ON TRIGGERS
Source: Notion
What’s Happening
Notion launched Custom Agents: autonomous “teammates” that can run on schedules or triggers and operate across Notion plus connected tools like Slack, Gmail, Figma, and Linear. Notion positions them as always-on workflow automation rather than a chat feature.
They are free to try for a limited period, then move to a credit-based model starting May 4.
Why It Matters
Agents are becoming admin defaults. Status reports, triage, routing, and internal Q&A can run without someone remembering to ask.
Pricing shifts to outcomes. If credits map to agent runs, teams will measure value more directly.
PERPLEXITY “COMPUTER” GOES MULTI-AGENT, IN THE CLOUD
Source: Perplexity
What’s Happening
Perplexity launched “Computer,” an AI agent product designed to break a goal into subtasks and assign work across multiple models and sub-agents. It is positioned as a managed alternative to DIY agent setups, with more structure and integrations.
Coverage highlights multi-model orchestration and longer-running workflows.
Why It Matters
The agent interface is converging. Fewer “chatbots,” more systems that plan, delegate, and execute.
Managed beats powerful for most teams. The product that wins is the one you can trust with credentials and workflows.
THE BIGGER PICTURE
Two forces are colliding: governments want AI capability without private vetoes, and labs want safety boundaries that don’t evaporate under pressure. That tension is now out in the open.
At the same time, agent products are moving from “cool demo” to “daily workflow.” The biggest winners will be the tools that make automation feel boring, predictable, and safe enough to hand over real access.
If this issue helped you make sense of AI’s chaos, forward it to a friend who shouldn’t be sleeping on this.
What did you think of today's edition?
Until next time,
Long Live AI