TOGETHER WITH

Good morning, human brains. Welcome back to your daily munch of AI news.

Here’s what’s on the menu today:

• New White House AI rules 🏛️

  The top AI firms voluntarily agree to invest in cybersecurity and more.

• Military funds AI chips with brain cells 🧠

  The Aussie government cuts a check to DishBrain, an AI chip with human brain cells.

• OpenAI’s Trust and Safety Leader is out 🏃‍♂️🏢

  Dave Willner, OpenAI’s Head of Trust and Safety has stepped down.

 Sponsor Bot Eat Brain | New here? Subscribe!

APPETIZER

The White House’s AI regulations 🏛️

The White House brokered an agreement between the top AI firms to commit to AI safety. Specifically: OpenAI, Google, Meta, Amazon, Microsoft, Anthropic, Inflection.

Huh, no Stability AI?

The agreement is “voluntary.“ Which means the punishment for breaking it? Yup, nothing.

Nice.

All 7 companies agreed to invest in cybersecurity, discrimination research, and watermarking systems for AI content.

We pinky swear.

There will be an official signing of the agreement at The White House on Friday.

Our take: This feels more like PR than anything truly actionable. Is it anything serious if there’s no oversight and no penalties for straying from the agreement?

BUZZWORD OF THE DAY

FROM OUR PARTNERS

The ChatGPT for presenting ideas beautifully.

The fastest and easiest way to create beautiful presentations and slide decks in minutes.

Gamma helps those with limited time and design skills turn messy, complex ideas into engaging, shareable content.

It’s a powerful web application, so there’s nothing to download.

Everything created is mobile-responsive and interactive by default.

Free to use, with an option to pay for unlimited AI credits.

MAIN COURSE

Military funds AI chips with brain cells 🧠

A chip/brain cell hybrid called DishBrain has gotten the green light.

It’s brain in a dish… Ha! Get it?

The team behind it are using the project to better understand how human brains might merge with AI. Yep, it’s using real human brain cells.

Cortical Labs and scientists from Monash University created DishBrain and taught it to play pong. Took all of five minutes, apparently.

But… to put that into perspective, biomedical engineer Thomas DeMarse got a petri dish of rat brain cells to fly a simulated F-22 — almost 20 years ago.

Back to Dishbrain: they just got $407,000 (in US dollars) from the Australian Defense Force.

The grant came from the NISDRG program (National Intelligence and Security Discovery Research Grants).

Now say that 10 times, fast.

The goal? Develop DishBrain chips to outperform current hardware and learn over a lifetime.

Razi claims DishBrain could improve strategy, robotics, automation, drug discovery, brain-machine interfaces, and more.

Our take: With the AI vacuum cleaner sucking in billions of dollars as we speak, $407,000 is a drop in the bucket. And Dishbrain can play pong, but getting it to perform Razi’s feels like a long ways away. Still: we’re seeing neurons interfacing with silicon at an accelerated pace, this time even with human neurons. And extra points for the knee-slap-inducing moniker.

A LITTLE SOMETHING EXTRA

OpenAI’s Dave Willner quits 🏃‍♂️🏢

OpenAI’s Safety and Trust Leader has stepped down.

Dave shared it’s to spend more time with his family.

Willner was the Trust & Safety Leader for OpenAI for a year and a half. Before that, he led Facebook and AirBnB’s trust and safety teams.

OpenAI’s looking for his replacement.

Maybe you?

Our take: Is this the real reason behind his departure? 🧐 We can only speculate for now.

MEMES FOR DESSERT

YOUR DAILY MUNCH

Think Pieces

How Marc Andreessen uses ChatGPT. Spoiler alert: he uses it to teach his kid about life. He claims AI is a lifelong ally, not a threat.

OpenAI’s statement comments on its agreement with The White House. More details on how they’re “moving AI governance forward.“

Is Meta’s Llama 2 truly “open-source?“ According to The Open Source Initiative, the real answer is “no.“

Startup News

Twitter is now X. Why X? The CEO, Linda Yaccarino, explains “X is the future state of interactivity.“

Stability releases FreeWilly. It’s a new open-access LLM that they claim has “exceptional“ performance.

LangChain reveals LangSmith. It’s a tool for debugging, testing, evaluating, and monitoring LLMs.

Research

Tackling the issue of slow sampling in diffusion models. This explains why it might happen and some potential fixes.

BoxDiff, a new method for text-to-image synthesis. This approach requires no training and gives users precise control over object and context synthesis.

CopyRNeRF. A new method to safeguard a NeRF’s (Neural Radiance Fields) intellectual property.

Tools

Gamma.ai [Sponsored] — create beautiful presentations quickly and easily with AI.

Airstrip AI — AI-powered legal assistant for businesses of all sizes.

Chitchat — a user-friendly LLM chat front-end to test out different LLMs.

Neum AI — connect your own data and keep your AI up-to-date.

TWEET OF THE DAY

If you didn’t already notice or hear about it… Yeah, Musk made it official.

Tag us on Twitter @BotEatBrain for a chance to be featured here tomorrow.

RECOMMENDED READING

If you like Bot Eat Brain, you might like these other newsletters too:

👨 The Average Joe — Market insights, trends and analysis to help you become a better investor. We like their easy-to-read articles that cut right to the meaty bits.

🕗️ 8AM Newsletter — 8AM delivers just one story, idea, or gem of advice for creators, entrepreneurs, and hustlers every morning at 8AM (ish).

💸 Who Sponsors Stuff — Find advertisers for your newsletter. Each issue includes 10 sponsors buying ads in newsletters right now.

AI ART-SHOW

Until next time 🤖😋🧠