Get ready for social media's user experiences and business models to reshape the chatbot-dominated AI world.

The big picture: New moves by OpenAI and Meta could add invasive advertising to AI's already-long list of problems, and the tech's power of persuasion and eagerness to please might supercharge the attention-grabbing capability of our social feeds.

Driving the news: OpenAI yesterday debuted Sora, an iOS app that combines an improved AI video engine with social features, including video sharing, scrolling through a feed and remixing.

• Meta last week introduced Vibes, its own AI-video app, and plans to start serving up content and advertising across its services based on the interactions people have with the Meta AI chatbot.
• The company announced those plans today, and said it will send notifications and emails on Oct. 7, with AI-driven ads and experiences beginning in December.

How it works: The Sora app, which is invite-only for now, includes a "cameo" tool that lets you add yourself and your friends to AI videos.

• As part of the sign-up process, users are required to complete a short live video repeating certain numbers or phrases and turning their head in particular ways.
• This both generates a cameo to authenticate a user's likeness and helps stop impersonation. Users can approve or remove videos made with their likeness.

What they're saying: OpenAI took pains to contrast what it's doing with Sora from past social media. Sora is optimized "for long-term user satisfaction," Altman said.

• "The majority of users, looking back on the past 6 months, should feel that their life is better for using Sora [than] it would have been if they hadn't," Altman wrote on his blog.
• "If that's not the case, we will make significant changes (and if we can't fix it, we would discontinue offering the service)."

Between the lines: People are already sharing all kinds of information with their chatbot of choice, from inner thoughts to highly personal medical data.

• These details could arm the makers of AI engines with all they need to deliver incredibly targeted advertising as well as content designed to keep people scrolling.
• OpenAI denies any current plans to put advertising in Sora.
• "A lot of problems with other apps stem from the monetization model incentivizing decisions that are at odds with user wellbeing," the company said in a blog post yesterday.
• "Transparently, our only current plan is to eventually give users the option to pay some amount to generate an extra video if there's too much demand relative to available compute. As the app evolves, we will openly communicate any changes in our approach here, while continuing to keep user wellbeing as our main goal."

Meta, meanwhile, has made clear its desire to merge AI with its proven advertising-based business model.

• The company has long reserved the right to use nearly any data shared with Meta AI however it sees fit, including to serve up ads or hone its AI models.
• The new video-sharing app and move to use Meta AI data for ads merely confirm it is putting those plans in action.

Yes, but: Even as it decries the past, OpenAI has made moves that could lead it down a similar path, especially as it looks to offset the massive cost of delivering generative AI video.

• OpenAI also recently added a "Pulse" feature to ChatGPT that proactively serves up information based on a person's interests. In doing so, it created a canvas that industry watchers noted seems tailor-made for advertising.

What we're watching: Many of the videos trending in Sora feeds featured Sam Altman, who enabled anyone to use him as a cameo.

• Popular videos included Altman trying to steal GPUs from Target and another of him outside Nvidia offices screaming for them to invest billions more.

As Washington races ahead with a hands-off approach to AI, much of the rest of the world is slowing down to set rules.

Why it matters: Companies that work across the globe will be dealing with different regimes, compliance costs and expectations — and the U.S. could get left out of this AI conversation.

Driving the news: That was made clear at last week's UN General Assembly in New York City, where the first Global Dialogue on AI Governance took place.

• Goals for this "global dialogue" are to align rules "to help build safe, secure and trustworthy AI systems," per a speech by UN Secretary-General António Guterres.

Threat level: If there are AI disasters in the future, the U.S. may not be part of any global agreements on how to mitigate or deal with them.

Zoom in: The Trump administration has its own ideas for how to best control and deploy AI.

• "We totally reject all efforts by international bodies to assert centralized control and global governance of AI," Office of Science and Technology Policy director Michael Kratsios said in his remarks to the UN debate on global AI.

Policymakers from Finland, Singapore and India — along with participants of AI safety institutes from Canada, China, the OECD and Singapore — were among the panelists at AI Safety Connect last week.

• No U.S. officials spoke at the event.
• AI Safety Connect, held at UNGA, was meant to spur discussion on what "red lines" for AI should look like globally.

What they're saying: "I would want [the U.S. government] to be more publicly supportive, and I wish that we could actually have a quicker move towards global governance of AI regime with full U.S. support," Nicolas Miailhe, co-founder of AI Safety Connect and founder of Paris-based startup PRISM Eval, told Axios.

• Uma Kalkar, Miailhe's chief of staff, told Axios: "We've had U.S. presence [at our events], not necessarily always government presence."
• Kalkar: "It's not something that's being ignored, and it's not something that's being sidelined. ... It's about who's ready to have those conversations in those specific multilateral spaces."

What we're watching: World leaders have gathered for the last two years to discuss AI governance at summits in Paris and outside London.

• Vice President JD Vance attended this year's summit, telling the world that "the AI future is not going to be won by hand-wringing about safety."
• India will host the next global AI summit in New Delhi in February, and how the U.S. approaches that forum will send a major signal to the international community.

• President Trump ordered $50 million for AI-driven pediatric cancer research, even as his administration cuts broader biomedical funding and pauses grants. (Axios)
• AI's impact on U.S. jobs is little different from the effect of previous waves of new tech like the personal computer and the internet, a study from Yale and Brookings found. (Financial Times)

Just setting up my Sora.