🍭 Trump Killed AI Rules and Now He Wants More 🔄

Good morning. The same people who said AI regulation was cringe are now comparing model releases to FDA drug trials.

Character development. Beautiful stuff.

Let’s dive in 👇

🍭 What’s Cookin’:

• Trump wants an FDA for AI

• Apple + Intel strike a chip deal

• Google drops Gemini Flash-Lite

AI Regulation
📜 Trump Wants An FDA For AI

The Bite:

The Trump administration is considering an executive order that would create a government vetting process for frontier AI models before public release.

National Economic Council Director Kevin Hassett compared the proposed system to FDA drug approval in a Fox Business appearance on May 7.

The discussions were sparked by Anthropic's Mythos model, which autonomously identified thousands of zero-day security vulnerabilities during testing.

Hours after Hassett's remarks, White House Chief of Staff Susie Wiles posted on X that the administration is "not in the business of picking winners and losers," and a White House spokesperson called the reporting "speculation."

Google, Microsoft, and xAI have separately agreed to share models with the National Institute of Standards and Technology (NIST) for voluntary pre-release testing.

Snacks:

• Hassett said models should be "released in the wild after they've been proven safe, just like an FDA drug"

• The White House revoked Biden's AI safety executive order within hours of taking office in January 2025

• Former AI czar David Sacks left the White House earlier this year; Treasury Secretary Bessent and VP Vance are now leading AI discussions

• Anthropic CEO Dario Amodei met with White House officials in mid-April, weeks after Trump barred federal agencies from using Anthropic's technology

• 80% of U.S. adults said they were concerned about AI in a late-March Quinnipiac poll

• The Cato Institute warned the proposal could give federal officials "a kill switch over speech and innovation"

Why it Bites:

This is the administration that killed Biden's AI executive order on day one, called safety regulation "fear-mongering," and sent David Sacks out to tell the world that America would win AI by building, not regulating.

Sixteen months later, one model from Anthropic found enough zero-days to make the Treasury Secretary go on television and talk about national security.

The power shift is subtle but significant.

The people now driving AI policy inside this White House (Bessent, Vance, Hassett) are not the Silicon Valley libertarians who set the original posture.

Sacks is gone. So the tech industry's assumption that this administration would stay permanently hands-off just lost its main character.

And the FDA analogy, even if it gets walked back, puts a frame on the conversation that's very hard to un-frame.

Once a senior official says "just like a drug," every future model release gets measured against that standard in public discourse.

The practical question is whether this lands as mandatory or voluntary.

Right now, NIST already tests models from Anthropic and OpenAI.
Adding Google, Microsoft, and xAI to that list is incremental, not revolutionary.

But an executive order would turn a handshake into a gate, with gatekeepers.

In summary, the same administration that wanted to preempt state AI laws may end up creating…

…the federal bottleneck that makes state laws look lightweight by comparison.

Steal This Prompt
🪵 Tiny Wooden Freaks With Main Character Energy

Turn any character concept into a handcrafted wooden figure that looks like it escaped a cursed artisan toy shop.

Use it to:

• Make fantasy characters feel collectible

• Turn brand mascots into wooden lore characters

• Create toy-style visuals with “Etsy but haunted” energy

Workflow:

1. Click this link (Prompt)

2. Paste into your AI model

3. Replace the #s with your character, pose, and wood tone

4. Watch it cook tiny wooden drama like Geppetto got VC funding

   Try the prompt on Snack Prompt

_ToolBox™
🧰 5 BRAND NEW AI LAUNCHES

🔮 Auvylo
Turn birth charts and Four Pillars into AI personas you can actually talk to — like therapy, but your ex’s emotional damage gets a dashboard.

📰 HeyNews
Trains on your newsletter archive and spits out drafts in your voice, so you can ghost blank pages instead of your subscribers.

🧑‍💻 display.dev
Publish AI-generated HTML or docs behind company login in one click — perfect for sharing agent-built prototypes without Slack chaos.

🧠 Vexilo
Claude Code power users: stop burning tokens guessing commands and start shipping like you secretly hired 31 AI interns.

🎥 knooth
Record your screen and let AI cut the awkward pauses, zoom the cursor, clean audio, and make you sound way more prepared than you were.

_{Everything Else}
🧠 You Need to Know

🤖 Google Ships Gemini 3.1 Flash-Lite For Agentic Workloads
→ The new efficiency-focused model delivers responses 2.5x faster at $0.25 per million input tokens, targeting high-volume agent tasks and translation.

⚡ Apple And Intel Reach Preliminary Chip Manufacturing Deal
→ Intel will manufacture Apple-designed processors in a foundry model, giving Apple a second source beyond TSMC and sending Intel shares up 15% on the news.

🔗 Grok Launches Deep App Integrations Via Connectors
→ SpaceXAI shipped Connectors for Grok, letting users wire in Gmail, Drive, Notion, GitHub, Linear, and other apps for read-write access directly inside the chatbot.

🏗️ Hut 8 Signs $9.8B AI Data Center Lease At Beacon Point
→ The 15-year, 352 MW deal could reach $25.1B with all renewal options exercised, bringing Hut 8's total contracted AI data center capacity to 597 MW.

🏛️ White House Considers Pre-Release Vetting For AI Models
→ The Trump administration is drafting an executive order to review frontier AI models before public release, with NEC Director Hassett comparing the process to FDA drug approval.