Good morning. It’s Thursday, July 16. SpaceX just did what few AI companies do after a privacy scandal: it open-sourced the tool. Days after researchers caught Grok Build silently uploading users’ entire codebases to the cloud, Musk’s company killed default data retention and put the code out in the open to rebuild trust.
Today we are covering:
SpaceX open sources Grok Build after the coding tool was caught uploading entire codebases to the cloud without consent
EU orders Google to open Android‘s AI features to rivals and share search data under the DMA
1Password lets Claude sign in to websites and complete tasks without ever seeing your passwords
Apple publishes ad policies for Maps, banning home services businesses that dominate Google‘s local ads
Bloomberg reports Apple is nearing an OLED iPad mini, with a revamped iPad Air and base iPad coming in 2027
Let’s dive in
SpaceX open-sources Grok Build after data retention furore
By Connor Jones via The Register
SpaceX open sourced the full codebase of Grok Build, its AI coding CLI, days after researchers at Cereblab found the tool uploading users’ entire repositories to Google Cloud storage without clear consent.
In one test, Grok Build sent 5.1 gigabytes of data for a task that needed just 192 kilobytes, roughly 26,000 times more than necessary, including full Git history with previously deleted secrets. SpaceX disabled default retention starting July 12 and says it is deleting all previously stored coding data.
The 844,530 lines of Rust code are now public for anyone to audit, a rare move for a commercial AI coding tool and a direct response to the trust deficit created by the leak.
𝕏: We’ve appreciated your feedback on X. Limits are reset, Grok 4.5 is available in the EU, and Grok Build is now open source. - Jason Ginsberg (@JasonBud)
EU forces Google to open Android to AI rivals and share search data
By Robert Hart via The Verge
The European Commission issued preliminary decisions requiring Google to open Android‘s AI features to competing assistants like ChatGPT and Claude, and to share anonymized search data with rivals under the Digital Markets Act.
Rivals must get free access to the same wake word activation, on-screen understanding, and background task execution that Gemini already uses on Android, while Google must share ranking, query, click, and view data from Google Search on fair terms.
Search data sharing starts January 2027, with Android interoperability changes reaching users by July 2027, giving Google roughly a year to comply or face further enforcement.
𝕏: Search Data to Rivals: Google’s Brutal EU Order Reshapes AI - Murtuza J Merchant (@murtuza_merc)
The best way to reach new readers is through word of mouth. If you click THIS LINK in your inbox, it’ll create an easy-to-send pre-written email you can just fire off to some friends.
1Password now lets Claude sign in to websites without seeing your passwords
By Zac Hall via 9to5Mac
1Password launched 1Password for Claude on Mac, letting Anthropic‘s AI agent complete browser tasks that require a login, such as booking travel, without the password or two-factor code ever entering Claude‘s memory or Anthropic‘s systems.
Claude requests only the specific login items a task needs, the user approves with a biometric prompt, and access lasts just for that session. A new Agentic Mode locks the vault so only approved credentials stay available while an AI agent controls the browser.
The feature is live now for Mac users on business, family, and individual plans, with support for payment cards and identity information planned later. It turns a partnership 1Password first outlined in March into a shipping product.
𝕏: BREAKING: 1Password launches browser integration for Claude, enabling the chatbot to access stored security credentials. Users can now leverage 1Password’s vault for Claude interactions. - Jan_RXed (@RXed_EU)
Apple quietly reveals how its Maps ads will differ from Google’s
By Sarah Perez via TechCrunch
Apple published the ad policies governing its upcoming Apple Maps advertising business, banning home services categories like plumbing, electrical, locksmith, and roofing that make up one of Google‘s largest local ad businesses.
The new rules, effective July 14, also ban cryptocurrency ATMs and bail bonds providers, and require case-by-case review for medical service ads. Apple says it will show only a single ad per search, marked with a small blue halo around the pin.
Apple has not set a launch date beyond “this summer” in the US and Canada, but the published rulebook signals the curated, navigation-first approach it wants Maps ads to take instead of mimicking a search engine like Google‘s.
𝕏: Did Apple just signal a third-party expansion of Apple Ads? Apple quietly updated its Advertising Services terms yesterday to allow for Apple Ads to operate on third-party surfaces, including non-Apple-owned apps and websites. This could be a signal that Apple intends to meaningfully expand Apple Ads beyond its owned apps. - Eric Seufert (@eric_seufert)
Apple Nears Launch of OLED iPad Mini; New iPad Air and Base iPad in 2027
By Mark Gurman via Bloomberg
Apple is preparing its biggest iPad mini overhaul in five years, adding an OLED screen for the first time along with a redesigned, more water-resistant chassis.
The new mini, code-named J510, uses a vibration-based speaker system to cut down on holes that let in water and dust, and could cost up to $100 more than the current model. The iPad Air will get its own OLED upgrade in early 2027, using a cheaper single-stack panel from Samsung.
The base iPad is not expected to move to OLED anytime soon, leaving Apple‘s cheapest tablet on older display technology while the rest of the lineup catches up to the iPhone and iPad Pro.
𝕏: NEW: Apple readies an end-to-end iPad overhaul, including a revamped iPad mini with OLED by October; a new entry level iPad early next year, a new iPad Air line in the spring alongside new iPad Pros and Apple Pencils. - Mark Gurman (@markgurman)
Trending in AI
Today’s big AI theme: AI’s growing pains are becoming impossible to ignore, from executive safety fears to child protection features to a fast-rising open-source challenger to Anthropic.
Meta now alerts parents if their teen discussed suicide or self-harm with its AI chatbot
Chinese AI start-up Moonshot to launch model challenging Anthropic’s lead
Nvidia reveals new AI model and expands Japan’s physical AI ecosystem
Thanks for reading to the bottom and soaking in our Newslit Daily fueled with highlights for your morning.
I hope you found it interesting and, needless to say, if you have any questions or feedback let me know by hitting reply.
Take care and see you tomorrow!
How was today’s email?
