Hey how’s it going?
I took a few days off (you noticed?) for Labor Day activities. Burgers, beers, a pool and plenty of college football. Hook ‘em. Not worried about it.
I also spent time building out a very AI-driven workflow for this newsletter. Nightly research across hundreds of sites, heat-index assessment, talking points, counter points, style guide application and more.
Of course, I’m still the inevitable author but it’s a big accelerator. I get to watch Alien Earth or whatever while most of the grind is taken care of for me. I’ll share a guide shortly … likely via zoom call … you interested?
Clay
Today’s AI News Snapshot
Score: 66/100 — Cautious Optimism: OpenAI pushed actually compelling safety features(!) and China’s new AI content labels (they’re kind of tricky) get mixed reviews. Plus, widespread gov adoption of Microsoft AI and a big OAuth breach doesn’t surprise us.
Positive Signal
There’s been a lot of controversy around AI usage and mental health this past month. A necessary conversation. Now, OpenAI says it will roll out Parental Controls within a month and route “sensitive conversations” to deeper reasoning models.
What I like about this news:
Expert informed by OpenAI’s own “Expert Council on Well-Being and AI” and a physician network of 250+ professionals.
Parental notifications when teens (the min. user age is 13) show signs of “acute distress” in their chats.
Part of a 120 day push so expect a sustained wellness awareness campaign by OpenAI into 2026.
What I have questions about…
That council is brand new (the first mention of it ever by OpenAI was just yesterday) and independence is unclear. Shouldn’t this be a shared body by the industry at large?
The ‘re-route’ is just to GPT-5 Reasoning so this is really just a move to ask the model to think a little more (vs. just optimize for cost) when the convo gets difficult. It’s a good step, absolutely. But more will be needed.
More positive signal:
Whoa. Honestly, whoa. China now requires visible and embedded labels on AI‑generated text, images, audio, video, and virtual scenes, with app stores expected to police AI features. From what I can tell, this applies to social media platforms for now. It’s the biggest real‑world test ever of watermarking the internet “at source.” Delicious, artisanal AI slop.
What I like about this news:
Rage against the deep-fake by getting us quickly out of the unlabeled deep-fake/misinformation anxiety era, and maybe solve for media literacy.
Promote human creativity by elevating content that is human (by it’s no-label prestige), and maybe even creating an environment that rewards it.
Force the conversation on how we should deal AI generated content globally. Should US regulators and other western bodies adopt similar? Will the platforms be compelled to do it themselves?
What I have questions about…
How the ?*#% are you going to label AI-generated Text and other content consistently? I kind of get it with images and video, but text? This probably comes down to “AI-supported creation” vs. “Fully-AI created” and how we label each.
Can’t replace wider training, and if you’ve been on X or FB lately, you know that obvious AI doesn’t keep people from sharing or being persuaded by misinformation. Research here agrees.
Chaos from fractured global implementation of these rules. Global companies will be forced to have separate features for users under this policy or implement these regulations into the global user-experience.
Everything Else
Major Opinions
The Washington Post argued AI competition is already curbing Google’s power, which reframes antitrust as a product race
Axios told readers to obsess over AI now, because — of course — the change is economy‑wide and coming fast. Interesting counter to all of the “AI winter” nonsense last week.
Government & Policy
Texas SB 20 took effect, banning obscene visual material that appears to depict minors. Not exclusively an AI thing, but directly targets AI-content.
Indiana University rolls out nation’s second-largest ChatGPT Edu access program for students.
Product Launches
OpenAI acquired Statsig, making Vijaye Raji CTO of Applications (!) After the Statsig acquisition and news of a CTO of Applications, OpenAI signals a quickening of focus on building B2B apps that integrate AI for more industry-specific use-cases. We’re going beyond ChatGPT.
I'd love to hear from you!
Thanks as always, human.
