😺 AI goes awry

PLUS: Buzzfeed's travel "guides"

Welcome, humans.

Elon Musk just became Twitter's most followed person, which isn't surprising since he owns the damn thing. Next up: the CCP becomes the most followed account on TikTok...

Here’s what’s going on in the world of AI today:

🪜 Midjourney takes a step back
🧑‍🍳 Buzzfeed churns out AI articles
💻 Claude + Slack integration
🤠 Friday Fun Fact

Midjourney Pauses Free Trials Amid Fake Photos 📸

Midjourney, the platform behind the fake (and hilarious) photos of the Pope and Trump, is pausing free trials amid exploits to its system.

The company's CEO announced on its Discord that tens of thousands of people created "throwaway" accounts to create free generative AI images. It was so bad that some paid users couldn't even get the service to load.

“We think the culprit was probably a viral how-to video in China. This happened at the same time as a temporary gpu shortage. The two things came together and it was bringing down the service for paid users.”

Midjourney CEO David Holz

Now, the company is disabling free trials and taking time to:

Work towards an AI "sus" detector to improve moderation.
Determine rules around controversial prompts like "arrest photos".
Have broader conversations on image moderation.

... all in real time.

Which got us thinking, Elon might have a point: AI is moving fast. Perhaps too fast for anyone to contain.

Zoom out: With AI, the mantra of "move fast and break things" has taken on a whole new meaning. Brace yourself for AI companies rapidly shipping products and then pulling back when things go awry.

This Startup Uses AI To Deliver Marketing Campaigns 👇

Most companies market products in a way they think might land with customers.

But RAD AI leverages proprietary AI to craft marketing messages they know will land with customers.

The B2B SaaS startup scans millions of web pages, Reddit threads, and APIs to craft a content strategy that taps into consumer psychology.
Their results are rad and the startup has landed clients like Accenture, Ro, MGM and Dignity Health.
Featured on Benzinga, Tech Crunch, Entrepreneur and Vanity Fair

After growing 310% last year with a $2.2 million run rate, RAD is raising a community round with thousands of investors participating.

Join over 3500 investors (VC and private) and invest on Wefunder before the round closes.

BuzzFeed Churns Out Travel Guides Using AI ✈️

Mr. Buzzfeed Worldwide.

The company recently released 44 travel guides on destinations worldwide in a matter of days. The kicker? AI helped them do it.

More specifically, Buzzy the Robot, a new creative AI assistant, helped Buzzfeed transform dozens of travel questionnaires into full-blown SEO articles.

An admitted experiment for the media company, the results were underwhelming: Buzzy had a knack for calling every destination a "hidden gem," among other irregular writing styles.

Why it matters: The experiment hints at a future potentially spammed with mediocre SEO articles like these.

On the other hand, ChatGPT might end up rendering BuzzFeed useless anyways. We personally wouldn't be upset if that happened!

Around the Horn 🦄

Anthropic announced that its Claude chatbot is now integrated into Slack workspaces.
The Washington Post is assembling a task force to study how AI will affect the paper's journalism.
Google's Bard got the majority of math questions incorrect on the SAT and similarly struggled on the writing section.
Eliezer Yudkowsky appeared on the Lex Fridman pod to talk about the dangers of AI.

Treats To Try 🍿

Featured: Flexible, scalable, affordable inference. Get 4x more inferences per dollar with Salad’s Inference Endpoints API. Start generating results in 2 minutes with Salad.
Create brandable and memorable domain names with Domains GPT.
Opus AI turns text into 3D metaverses, games, and movies.
Akkio's Chat Explore can turn your data into charts, analyses, and explanations.
YakGPT is a ChatGPT UI that runs locally on your browser and includes mic integration.
My AI Front Desk: AI-powered receptionist that you can call and text.
Socket AI uses ChatGPT to examine npm and PyPI packages for potential security issues.

DM us links on Twitter: @nonmayorpete & @noahedelman02.