AI Briefing: As tech giants add more AI tools, Runway and Getty Images team up

Getty’s content library provides training data for a new enterprise-focused AI model for generating AI videos. The new model will also enable clients to fine-tune it with their own own data sets. The partnership is “the next stepping stone for both commercial adoption,” said Runway co-founder and CEO Cristóbal Valenzuela, who added that being able to fine-tune the model with other data sets is “where the real magic starts to happen and something we haven’t seen so far.”

There are differences between Runway’s approach to its current generative AI model and its new “Runway <> Getty Images Model (RGM)” which is still in beta with a broader release planned in upcoming months. For example, Valenzuela said the RGM is “highly creative and highly curated” and also legally compliant thanks to the data being pre-licensed by Getty. It will also let movie studios, advertisers and broadcasters “ingest videos, references, or shows or series you own, or products we’re working on that have never seen the light [of day] and have never been public.”

“I think a lot of the AI models we’re using right now are general-purpose models that are made for all sorts of things, but things that they’re not really excelling at — like specific domains or verticals,” Valenzuela said. “For artistic practices, for creatives and for filmmakers, control is key. This is where these models really come in.”

The partnership comes not long after Getty Images released its own AI image platform in September. According to Getty Images chief product officer Grant Farhall, the company’s customers want to use more video content, but it can still be time-consuming. However, he said Runway’s tools help not just to generate content but also provide ways to move in various directions, pan, zoom in and out, and other tools. 

This doesn’t mean Getty thinks AI videos will totally replace human-made ones. Instead, Farhall thinks it gives people options for what they need, including possibly human-created video.

“When you are seeking authenticity, you can do that from a pre-shot library,” Farhall said. “AI is exciting where you can stretch your imagination, where you can create something that would be either impossible or difficult or really time-consuming to shoot. Again, the right tool for the right job, but certainly AI allows for more efficient creation of things that would be really, really hard.”

AI news

• Last week, European Union officials continued debates — and reached a deal — on the proposed EU AI Act, which would provide a range of new AI regulations for the 27 countries that make up the union. 

• British regulators have begun to gather information to see if Microsoft’s partnership with OpenAI violates antitrust laws. On Friday, the UK’s competition watchdog, the Competition Marketers Authority, issued a statement inviting commentary about whether Microsoft and OpenAI’s ties have “resulted in a relevant merger situation and, if so, the impact that the merger could have on competition in the UK.”  

• Several social platforms came under scrutiny last week for various uses of generative AI. Meta faced criticism for training its AI platform with users’ photos from Facebook and Instagram; TikTok faced problems with deepfake videos of Hamas victims showed up in various posts; and Alibaba was called out for training its AI video model using videos from famous TikTok creators.

• OpenAI said it’s investigating concerns about ChatGPT “getting lazier,” following reports of users finding the chatbot becoming less predictable.

Products and prompts: Other AI-related announcements

• One of the biggest announcements last week was Google’s debut of Gemini, a powerful new AI model that aims to compete with OpenAI’s GPT-4 and other large language models.

• Meta released new updates to its own AI tools across Facebook, Instagram, Messenger and WhatsApp, including nearly two dozen ways to use generative AI for search, social discovery, advertising and business messaging.

• To tout its own AI offerings, SAP Business released a new ad campaign starring an AI-animated cat painted in the style of Vincent Van Gogh.

• Just in time for the holidays, the sleep and meditation app Calm used AI to recreate the voice of James Stewart, the actor who played George Bailey in “It’s A Wonderful Life.” Created in collaboration with the AI voice startup Respeecher — and with consent from Stewart’s family — Calm’s version of the actor’s voice reads a new 45-minute story called “It’s A Wonderful Sleep Story.“

• McDonald’s announced a new AI partnership with Google, which will reportedly include a new “Ask Pickles” chatbot to help workers with daily operations.
  
• Creator Karen X Cheng posted her keynote speech from October’s Adobe MAX conference — titled “The Artist vs The Algorithm — which provides a candid view of the challenges creators face across various social platforms.

Other AI news from Digiday

• Rapid developments in artificial intelligence have led to new AI names entering the cultural vernacular for everything from new startups and chatbots to large language models and AI products. (Digiday’s search of the U.S. Patent Office’s trademark database also discovered that OpenAI filed new trademark application for GPT-6 and GPT-7 back in October.)

• Lexicon founder and CEO David Placek spoke with Digiday about AI naming trends, what makes a good name and how his agency — which came up with iconic names like Blackberry and Azure — is approaching naming projects for various new AI startups.

• An AI task force for publishers is expanding and evolving to include more people within various member companies.