OpenAI is unveiling today its long-expected teen experience for ChatGPT that features parental controls, efforts to encourage real-world relationships and a focus on learning rather than serving up answers.

Why it matters: The move is likely to rekindle debate over whether teens should be using chatbots at all.

Driving the news: OpenAI says ChatGPT for Teens will differ in significant ways from its standard offering. Among the differences:

• Learning over answers: The teen version will steer users toward working through schoolwork rather than simply providing answers, including a Study Mode and new "responsible homework reminders" that can detect when a teen appears to be trying to shortcut an assignment. Parents and teens can also set "Study Hours" when Study Mode is on by default.
• Stronger content safeguards: Teen accounts will have additional protections around self-harm, eating disorders, violence, dangerous activities and sexually explicit or graphic content. OpenAI is also expanding parental safety notifications to include certain eating-disorder-related situations.
• Less human-like relationships: ChatGPT will be barred from using romantic language or terms of endearment with teens and will be more strongly instructed not to suggest that it has feelings, consciousness or emotions. OpenAI says the changes are intended to reduce anthropomorphism and emotional dependence.
• More nudges to step away: Teens will get more frequent break reminders during extended use, including reminders that they are interacting with AI, as well as warnings before uploading potentially private or sensitive images.

How it works: Users who say they are 13 to 17 — as well as those OpenAI's age-prediction system estimates are under 18 — will automatically be placed into the teen experience.

What they're saying: OpenAI says teens are accessing AI already.

• "If you don't build a safe space for them, they're going to go to a less safe model experience," said Lauren Jonas, OpenAI's head of youth and families.
• "We believe that if you're going to have these tools, they need to be by default safe," she said. "But then it's also critical that teens have safe spaces to engage with these new emerging technologies."

The other side: Some say kids and younger teens shouldn't be building relationships with AI systems at all.

• "Social AI companions are not safe for kids," Jim Steyer, founder and CEO of Common Sense Media, said in a statement last year. "They are designed to create emotional attachment and dependency, which is particularly concerning for developing adolescent brains."

OpenAI faces legal action over its interactions with young people, including a prominent lawsuit from the family of a 16-year-old who killed himself following lengthy interactions with ChatGPT.

• The company pledged last year to increase safeguards for younger users and those in crisis. It added parental controls in September.

The big picture: Teens are already using AI in large numbers, with a Common Sense Media study last year finding nearly three-quarters of U.S. teens saying they use AI for companionship — and more than half of those doing so at least a few times a month.

• More kids will also soon have access via their schools, as Google last week turned on its Gemini chatbot within Google Classroom for K-12 schools.

Anthropic's second quarter revenue exploded from a year ago, helping push the AI lab's annualized revenue run rate above $65 billion ahead of its expected IPO, according to figures reported by Bloomberg.

Why it matters: By its latest reported revenue run rates, Anthropic has pulled ahead of rival OpenAI.

By the numbers: Anthropic generated more than $11.5 billion in preliminary revenue in the second quarter, according to documents seen by Bloomberg.

• That's more than 14 times what it generated in the same quarter last year and more than double Q1 revenue of $4.73 billion, a sequential increase of more than 140%.
• The reported $65 billion run rate — an estimate of future annual revenue based on a short period of data — was hit at the end of July, per Bloomberg, and implies a sevenfold increase from where it was at the end of last year.

Between the lines: That growth is particularly notable given the scale Anthropic had already reached entering the quarter, according to an Anthropic investor who spoke with Axios.

OpenAI's latest revenue run rate hit $40 billion, according to a message shared internally by co-founder Greg Brockman last week.

• Caveat: The two companies may not measure revenue the same way.

Follow the money: Anthropic could go public before its rival.

• The company is meeting with potential new investors ahead of its planned IPO, expected in September or October, according to a source familiar with the matter.
• Morgan Stanley, Goldman Sachs and JPMorgan are working with Anthropic on the offering.
• IPOs would help the companies tap additional pools of capital to secure more capacity for model usage — and going first could give the Claude maker a leg up.

What we're watching: Whether efficiency becomes the next battleground between the leading AI labs as they look to expand margins.

• "Anthropic was much more token efficient than OpenAI but OAI has closed some of the gap," Gavin Baker, managing partner at Atreides Management, told Axios' Dan Primack via text.
• OpenAI and Anthropic have both cut deals with major inference providers and have their own chip ambitions, which could allow them to lower the cost of servicing AI queries going forward.

Keep reading.

• BNY CEO Robin Vince welcomed his 2026 analyst class with a nudge toward AI usage. (Axios)
• OpenAI announced a massive data center in Ohio that will be guaranteed by Nvidia. (Axios)
• Google could get access to a deluge of Spirit Airlines data for just $10 million. (Axios)

Mady here with some admiration for the 404 Media team's coverage of AI labs buying, scanning then destroying old books.

• Using an AirTag, the team found that a shipment of rare books was sent to an Amazon facility that processes physical books for LLM training, which 404 found involves destruction of these books.
• Amazon, of course, started out a book seller.