The Prompt: Anti-AI Pledges Gain Popularity

The Prompt is a weekly rundown of AI’s buzziest startups, biggest breakthroughs, and business deals. To get it in your inbox, subscribe here.

Welcome back to The Prompt.

As companies double down on their AI strategy, some firms are taking a different approach by making “anti-AI pledges,” vowing to stay away from generative AI altogether. Over the weekend, illustration app Procreate’s CEO Jame Cuda said on X that the company will not be introducing any AI features into its product. “I don’t like what’s happening to the industry, and I don’t like what it’s doing to artists,” he said. The announcement garnered widespread praise among creatives who have long voiced concerns about their content being scraped without consent and compensation to train generative AI models.

Other brands have made similar proclamations, in part for publicity. Cara, a social media platform for artists, prohibits people from uploading AI-generated artwork until more stringent laws for data practices are introduced. Dove has pledged to use real women in its ads instead of AI models and Discover has similarly advertised that its call center is only staffed by humans and not robots.

Now let’s get into the headlines.

ETHICS + LAW

San Francisco City Attorney David Chiu’s office filed a lawsuit against the owners of 16 websites that use AI to generate nonconsensual pornographic images of women and girls. These AI-powered “undressing” apps were collectively visited 200 million times in the first six months of 2024, according to the lawsuit. “What we know is these images are used to bully, humiliate, and threaten women and children,” Chiu told Forbes. “We look forward to seeing who comes out of the darkest corners of the internet to defend their actions.”

Plus, Anthropic was hit with a class action lawsuit by a group of authors who allege that the AI company downloaded pirated copies of their copyrighted books and used them to train its large language models. While OpenAI and Microsoft are already facing a slew of copyright infringement lawsuits from fiction and nonfiction writers, this is the first time Anthropic is being sued by authors. “Anthropic’s model seeks to profit from strip-mining the human expression and ingenuity behind… those works,” the lawsuit says.

POLITICS + ELECTION

Former President Donald Trump posted AI-generated images of Taylor Swift fans wearing shirts with the slogan “Swifties for Trump.” On Truth Social and X, he also shared an AI-generated image of Taylor Swift in an Uncle Sam outfit, urging her fans to vote for him. Additional fake images shared by the former president falsely suggested Vice President Kamala Harris is a communist.

These political AI-generated images come shortly after Elon Musk’s AI startup xAI released a beta version of its Grok 2 model that has far fewer guardrails than other image generators. The AI model can churn out hyper realistic images of political leaders in different scenarios; examples created and shared by users included Barack Obama doing cocaine, Donald Trump standing with a pregnant woman that faintly resembled Kamala Harris and Trump firing two handguns.

AI DEAL OF THE WEEK

Conversational AI startup EliseAI raised $75 million in a Series D funding round at a $1 billion valuation. The company, cofounded by University of Cambridge graduates Minna Song and Tony Stoyanov, develops AI chat assistants that residential property managers use to communicate with renters for things like lease renewals, booking apartment tours and maintenance requests.

DEEP DIVE

Rather than train large scale, compute-intensive AI models that can perform a variety of tasks, AI companies are increasingly shifting toward smaller-size models that can carry out specialized functions and do them well. Enterprise-focused generative AI startup Writer, for instance, launched two new models that specifically target medical and financial use cases.

The AI models can then be used to build applications to automate processes or what’s called “workflows” within companies. A wealth manager, for instance, can use Writer’s AI model, Palmyra Fin, to assess the market and find which stocks to drop or add to their portfolio to maximize profits, CEO May Habib told Forbes. The medical model can look through a press release for a cancer drug and evaluate the claims made in each sentence. “A year ago we could never do something like this,” she said.

Enterprises have been moving quickly to add AI features to their core products, but they’re not always successful. As the market gets crowded with AI models that have similar capabilities, “people are overwhelmed with choice,” Habib said. Domain specific models could be a way for companies to differentiate themselves, she said.

“We are in a nose dive to the trough of disillusionment because [enterprises] are trying to use the LLMs out of the box and what they need to be doing is really setting up microservices on top of the LLMs,” she said.

AI INDEX

$196 million

How much tech companies have spent this year on advertising their AI products on TV, according to The Washington Post.

QUIZ

This tech giant announced a new feature for its latest smartphone that allows the photographer to be added into group photos using multiple shots and AI.

1. Apple
2. Google
3. Samsung
4. Intel

Check if you got it right here.

MODEL BEHAVIOR

Cats, one of the internet’s most beloved animals, have become the latest muse for AI-generated images. AI-generated images and videos of cats of all shapes and sizes are spreading on Instagram, Youtube and TikTok. Some creators are even developing elaborate, and sometimes disturbing videos with an AI-generated cat as the protagonist. In one, an AI cat becomes a police officer and kills another cat to seek revenge; in another, a cat appears to eat the content of a dirty diaper, the Washington Post reported.