Welcome, humans.

George Carlin’s estate is suing the creators of an AI-generated Carlin video impression. This isn’t the first time Carlin’s comedy has ended up in court. Remember his 1972 monologue "Seven Words You Can Never Say on Television" that landed in the Supreme Court? (we’ll let you Google Perplexity the words yourself, cause, well, we’re not trying to receive the same blowback he did).

Here’s what you need to know about AI today:

• Taylor Swift’s graphic AI photos have reignited concerns about AI deepfakes.

• ChatGPT is beta-ing a feature that enables access to all your GPTs in one chat.

• Midjourney is broadening web access and rolling out new features for its v6 iteration.

• ElevenLabs banned the creator of the Joe Biden deepfakes.

Taylor Swift’s AI deepfakes highlight a deep-rooted problem.

Last Wednesday, explicit AI deepfake images of Taylor Swift caused an uproar after going viral on X.

One post garnered over 45M views and stayed live on X for 17 hours (!) before being removed, causing enough of a stir to grab the attention of Microsoft’s CEO AND The Biden Administration. ↓

So WTF happened?!

A 404 Media investigation traced these images back to 4chan and Telegram group chats known for sharing non-consensual sexual imagery of women.

And while most major text-to-image tools like Microsoft have rules against such content, many loopholes exist, including intentionally misspellings celebrity names.

And don’t even get us started on X/Twitter/Elon Musk’s no-seatbelt rollercoaster park.

Even after taking forever to remove some posts, more images continued to spread on the site through alternate accounts. (Yesterday, X announced it’s hiring 100 new content moderation employees. Coincidence? I think not…).

While this incident is alarming, it's part of a much broader, troubling trend of non-consensual sexual AI deepfakes that has been emerging for years.

*more on this in a minute, but first…*

FROM OUR PARTNERS

Mixtral on OctoAI: 12x lower costs, 30% faster speeds vs GPT-4-Turbo

Some of the savviest builders in AI are ditching OpenAI for OctoML, and here’s why:

Building products on top of OpenAI can be expensive AF, and if you’re not throwing big bucks at them, latency and scale become problems. 

Enter OctoML—the fastest, most affordable option to run Mixtral 8x7B (an open-source model that outperforms GPT-3.5). 

Customers switching from GPT-4 Turbo see comparable quality, 12x savings, and 30% lower latency.

Try OctoAI Text Generation and snag $10 in free credits when you sign up today.

Graphic AI-generated deepfakes have *been* a massive problem.

Taylor Swift, despite her many "firsts," is not the first to fall prey to sexual AI deepfakes.

• WIRED found that >244,625 non-consensual deepfake porn videos have been uploaded in the past 7 years.

• And AI-generated porn photos have snowballed 290% since 2018, with many victims being teenage girls.

Side note: this is not even the first time T-Swift has been deepfaked—AI-generated Facebook ads depicted Swift endorsing Le Creuset Cookware a few weeks back:

These days, tools like Midjourney can now generate photos virtually indistinguishable to the human eye (& are only improving). If development accelerates without the right safeguards, these issues will only get worse.

Hate to be “I told you so” on this topic, but we’ve been sounding the alarms on AI deepfakes long before they sparked national outrage and became a hot topic in Congress…

Well that topic was gloomy…

To lift your spirits, we’re giving away $250 to one reader to buy any AI tool!

Just refer a friend to The Neuron using your unique link below.

It took us 7 seconds to share ours (seriously, trying beating that!).

Stay tuned—we're revealing the winner this Wednesday!

Around the Horn.

→ A new analysis found that the new Gemini Pro (which powers Bard) outperforms every model except ChatGPT-4 Turbo.

→ Check out this video on how a startup is using AI to insert product ads into videos.

→ Google debuted a new AI video generator called "Lumiere” last week.

→ The creator behind the Joe Biden deepfakes telling people not to vote was banned by ElevenLabs.

→ Discover how to move from AI hype to practical use cases that drive ROI at A.Team’s ROI of AI event (IRL & online).

Treat To Try.

OpenAI’s Cookbook is an invaluable resource for anyone working with OpenAI’s tools.

Dozens of guides and productivity hacks. ↓

We curated the top 27 tools you need for work in our “Top Tools For Business”!

Monday Meme.

A Cat's Commentary.