Welcome to The Week in Generative AI, a weekly column for marketers from Quad Insights that quickly sums up need-to-know developments surrounding this rapidly evolving technology.

Zoom faces criticism over AI training

Although Zoom updated the terms of service for its popular videoconferencing app in March, it took until this week for the internet hivemind to take the company to task over claims that those terms would allow Zoom to train its AI on user data.

Gizmodo’s Thomas Germain writes that “Zoom’s AI policies flew under the radar until a post on the subject sparked outrage on the influential forum Hacker News,” where a contributor flagged the lack of an opt-out feature for what’s known as “service generated data,” which includes a variety of data and other information Zoom collects by analyzing user content and behavior. (This data differs from “customer content,” which includes video, audio and chat transcripts.) Germain notes that “Although the policy literally says that Zoom reserves the right to train AI on your calls without your explicit permission, the Terms of Service now include an additional line which says, essentially, we promise not to do that.” (That additional line was added on Monday.)

Over at the Associated Press, Melissa Goldin clarifies a few points on the matter: “The current terms would not allow the company to tap user-generated content like video and chat for AI training without a customer opting in. However, once a meeting host agrees, other participants would have to leave if they don’t want to consent.”

Related coverage:
“Zoom can now train its A.I. using some customer data, according to updated terms” (CNBC)
“Zoom’s Updated Terms of Service Permit Training AI on User Content Without Opt-Out” (Stack Diary)
“Is Zoom using your meetings to train its AI?” (Vox)
“Zoom CEO admits mistake as terms-of-service changes raise AI fears” (Axios)

Deep(fake) cuts from Google and Universal Music?

Have you ever wanted to hear what it would be like if Taylor Swift sang your voicemail greeting? Or if you could have a personalized rap battle with Jay-Z? Well, thanks to the power of AI, that might become a reality.

According to the Financial Times, Google and Universal Music Group are in discussions to build an AI tool that will allow you to create AI-generated sounds using the music of major artists. The AI tool would reportedly be able to replicate artists’ voices with such precision that it will be almost impossible to tell the difference between the real thing and the AI-generated version.

“Under the licensing deal, the relevant copyright owners would be paid for the use of their likeness and would have the option to opt in to give UMG and Google permission to license AI-generated music using their voice,” writes Gizmodo’s Nikki Main. “Google and UMG are in the early stages of negotiations over creating the deepfake tool, and there aren’t currently any plans to immediately launch it.”

More Google AI news:
“Google Unveils New AI Tools To Help Advertisers This Holiday Season” (Search Engine Journal)
“Google says AI systems should be able to mine publishers’ work unless companies opt out” (The Guardian)
“Google Search’s Latest AI Feature Is a Grammar Checker. How to Use It” (CNET)

U.S. Government launches AI Cyber Challenge

The White House and the Defense Advanced Research Projects Agency (DARPA) this week announced a two-year competition, called the AI Cyber Challenge, that seeks to shore up AI cyber defense, with $18.5 million in prizes to be doled out to the top hackers. This partnership follows a summer of AI-related initiatives from the federal government, and the major players in AI are involved, as Anthropic, Google, Microsoft and OpenAI will make their models available for the competition. Reporting for Engadget, Andrew Tarantola writes that the “winning team will also be strongly encouraged to open-source their resulting program. The competition is bringing on the Open Source Security Foundation (OpenSSF), a Linux Foundation project, as an advisor to the challenge.”

Related coverage:
“DARPA launches two-year competition to build AI-powered cyber defenses” (TechCrunch)
“How Silicon Valley is helping the Pentagon in the AI arms race” (Financial Times)
“The AI crackdown is coming” (The Atlantic)
“Microsoft’s AI Red Team Has Already Made the Case for Itself” (Wired)

Further reading

“Exclusive poll: Americans distrust AI giants” (Axios)
“FEC could limit AI in political ads ahead of 2024” (The Verge)
“Disney explores cutting costs through AI use” (Ars Technica)
“How Mars, Colgate-Palmolive, Nestle & Coca-Cola are Exploring Generative AI” (Consumer Goods Technology)
“AI’s black box problem has clients cautious but curious, agency execs say” (Digiday)
“News outlets demand new rules for AI training data” (The Verge)
“Nvidia announces new chip to power AI models, reduce costs” (Fox Business)
“OpenAI adds ‘huge set’ of ChatGPT updates, including suggested prompts, multiple file uploads” (VentureBeat)

Thanks for reading. We’ll see you next week as the AI beat continues.

Previously: “The Week in Generative AI: August 4, 2023 edition”