- Horizon AI
- Posts
- Check This New Google Privacy Setting Before It Starts Collecting More of Your Data 🔒
Check This New Google Privacy Setting Before It Starts Collecting More of Your Data 🔒
OpenAI unveils its first custom AI chip 👀
Gina Acosta
June 25, 2026
In partnership with
Welcome to another edition of Horizon AI,
Google recently began rolling out a new privacy setting that many users have overlooked. In today's issue, we'll show you what it does and how to turn it off if you'd rather keep your data from being saved and used to improve Google's services and AI models.
Let’s jump into it!
Read Time: 4.5 min
Here's what's new today in the Horizon AI
OpenAI Introduces Its First Custom Chip, Developed with Broadcom
Report says Grok's Traffic Is Mostly Driven By Adult Content
AI Tutorial: A New Google Privacy Setting Is Rolling Out — Here's How to Turn It Off
AI Tools to check out
AI Findings/Resources
The Latest in AI and Tech 💡
AI News
OPENAI
OpenAI Introduces Its First Custom Chip, Developed with Broadcom
OpenAI has introduced Jalapeño, its first custom-built AI inference chip, designed to power its inference workloads more efficiently and reduce the company's reliance on Nvidia GPUs.
Details:
Jalapeño is purpose-built for inference, the process of running trained AI models to generate responses to user requests.
The “Intelligence Processor” was designed and manufactured in collaboration with Broadcom, with OpenAI stating that its own AI models assisted in the chip's development process.
While the chip remains in testing, OpenAI says early benchmarks show significantly better performance per watt than current state-of-the-art alternative.
OpenAI emphasized the chip's low operating cost when running real-time coding models, but more performance-intensive tasks, like pre-training, are still expected to rely largely on Nvidia hardware.
OpenAI described Jalapeño as the "first AI accelerator in a multi-generation compute platform the companies are building together to make advanced AI faster, more reliable, and more accessible to more people." The companies are aiming for initial deployment by the end of 2026, "expanding in the years ahead."
TOGETHER WITH CUEY
ChatGPT and Claude often disagree. Cuey shows you when.
You ask ChatGPT something important. It answers with total confidence. Ask Claude or Gemini the same thing, and the answer changes. One of them is wrong, and you are about to act on it.
Cuey runs your prompt past ChatGPT, Claude, and Gemini in a single tab, so you spot any disagreements before they cost you. Your memory and context follow you across all of them, no starting over. It works beside the tools you already use, with Grok, Mistral, and more than 30 others.
Free on Chrome, nothing to set up.
XAI
Report says Grok's Traffic Is Mostly Driven By Adult Content
A new report claims that Elon Musk's xAI is leaning into explicit content generation as a core driver of Grok's traffic, with NSFW content now accounting for the majority of the platform's activity.
Details:
According to The Information, two former xAI employees estimate that well over half of all Grok traffic goes to NSFW images, videos, roleplay chats, or other adult content. Even Grok's coding model sees frequent requests of this nature.
Vital Knowledge analyst Adam Crisafulli described the strategy as "a desperate attempt for relevancy," arguing that xAI has fallen further behind leading AI developers.
One notable controversy earlier this year involved Grok's image tools being used to "undress" photos of real people without their consent, forcing the company to introduce measures to prevent further misuse.
Grok was launched by Elon Musk as a direct competitor to ChatGPT, Claude, and Google Gemini, with a focus on being a less restrictive AI assistant. However, the platform has struggled to keep pace with leading frontier models, with its web traffic declining this year while competitors like Claude (+369%) and Gemini (+40%) have seen significant growth over the same period, according to Similarweb.
AI Tutorial
A New Google Privacy Setting Is Rolling Out — Here's How to Turn It Off
Did you get this email? Google started sending everyone this email last week about a new setting.
The new setting controls whether media used with supported Google search services (such as images submitted through Google Lens) will be used to improve the company's services and AI models. Google says the data eventually gets stripped of your identity once it's used for training, but they don't say when, and the collection is already happening.
How to turn it off:
On your phone or computer, visit: http://myactivity.google.com/activitycontrols
If you see a section called Search services history, tap into it and turn off “Save media.”
If that section isn't there yet (which is the case for most people right now, as it's rolling out gradually), go to Web and app activity and uncheck Voice, audio & visual search history as a stopgap.
You can then set a reminder on your phone to go back in three or four weeks and confirm the Search services history toggle is off once it appears on your account, as Google says it is "rolling out gradually over the next few months."
AI Tools to check out
🔥 Plotly Cloud: Transform any dataset into a data app in minutes.
✅ Exa Connect: Connect your agents to the world's public and private data.
📞 AgenticCalling: Give your AI the power to make phone calls.
👉 Propane: Automatic customer context for product teams and agents.
🌐 Aside: The first AI browser built to do real work for you. It works across your logged-in websites and handles complex work other agents can't.
AI Findings/Resources
💂 Why U.S. AI giants like Anthropic, OpenAI are launching major expansions in London
🗣️ OpenAI's upcoming Bidi 1 voice model will be able to translate in real time
The latest in AI and Tech
The company says the model is now better at understanding the intent behind a question, handles complex constraints more reliably, and delivers more useful shopping and local recommendations.
Top AI researchers Jonas Adler and Alexander Pritzel are leaving Google for Anthropic, according to Bloomberg.
Senior research engineer Arthur Conmy also announced his departure from Google to join Anthropic, "to work on aligning upcoming models as they're trained."
The design platform has unveiled a major update introducing code layers, built-in support for animations, and the ability to create custom plug-ins for various tasks using AI.
Google has integrated computer use directly into Gemini 3.5 Flash, allowing developers to build AI agents that can interact with websites, desktop apps, and mobile interfaces.
A coalition of nearly 400 local newspapers has filed a lawsuit against OpenAI and Microsoft, accusing the companies of “scraping, copying, and ingesting” their journalism without permission or compensation to train AI models.
The case adds to a growing wave of legal challenges against OpenAI, which already faces lawsuits from major outlets including The New York Times, Ziff Davis, Merriam-Webster, and Encyclopedia Britannica.
That’s a wrap!
We'd love to hear your thoughts on today's email!Your feedback helps us improve our content |
Not subscribed yet? Sign up here and send it to a colleague or friend!
See you in our next edition!
Gina 👩🏻💻