NEWS
Tech News
Sources Ars Technica CNET News Engadget Stratechery by Ben Thompson InfoQ MIT Technology Review Nature NYT > Technology Official Android Blog Slashdot TechCrunch The …
Sources
Story of the Day
OpenAI has signed a massive 10-year lease for a colossal 8-gigawatt Ohio data center built by SoftBank’s SB Energy, with Nvidia stepping in to guarantee up to $105 billion in lease and power obligations. Built on formerly federal land once used for uranium enrichment, the landmark project requires an unprecedented 10 gigawatts of new power generation and cements Nvidia’s deep financial entanglement with its largest infrastructure buyers.
Top Stories
Amazon’s AI Spree Exposed: AirTag Tracks Rare Books to Spines-Off Destruction · Ars Technica An investigative report confirmed long-standing industry rumors of destructive AI training practices by using a hidden AirTag to track a rare book directly to an Amazon warehouse in Las Vegas. At this specialized “VGT3” facility, employees systematically cut the bindings off bulk shipments of rare books to scan them rapidly for training data, destroying the physical copies in the process. Amazon deflected the findings, releasing a generic statement that it buys books commercially “to improve products and services” while refusing to comment on the literal destruction of rare literature.
Stripe Acquires AI Model Aggregator OpenRouter in $7 Billion Transaction · Bloomberg Stripe has agreed to acquire AI routing platform OpenRouter in a massive deal valued at more than $7 billion. OpenRouter allows developers to access and optimize over a hundred different AI models, a service that has become highly lucrative as companies look to drive down AI costs and move away from seat-based software models. The deal is an implicit bet on a multi-model future where aggregation layers hold the real power, shifting Stripe deeper into core AI developer infrastructure.
Anthropic’s Claude Watermarking Prompts User Backlash and Subscription Cancellations · Slashdot Anthropic confirmed it is rolling out an invisible “SynthID-Text” watermarking system for Claude to comply with the European Union’s landmark AI Act. However, professional users are panicking and canceling subscriptions because the watermark remains even when Claude merely proofreads, translates, or edits original human-written text. Freelance developers and consultants worry that these overzealous marks could raise questions about authorship, violate client agreements, or trigger contract penalties.
Meta Heads to Oakland Federal Court in $1.4 Trillion Child Addiction Trial · Bloomberg A high-stakes federal trial begins this Tuesday as dozens of US states accuse Meta of intentionally designing addictive features on Instagram and Facebook that harm young users. The coalition of state attorneys general also alleges Meta violated the Children’s Online Privacy Protection Act (COPPA) by knowingly collecting data on children under 13 without parental consent. While the statutory penalties theoretically reach a dizzying $1.4 trillion, the judge has called that figure unreasonable, but the final ruling could establish a precedent for thousands of similar pending lawsuits against the social media giant.
Bipartisan Rebellion Against Flock Safety’s Surveillance Cameras Triggers Local Cancellations · MIT Technology Review A bipartisan wave of US local governments has begun canceling contracts for Flock Safety’s license plate readers over civil liberties concerns, making July the highest month for cancellations since tracking began in 2021. Bipartisan local officials are pushing back against the mass surveillance dragnet, citing high costs, low utility, and a slew of stalking incidents where police officers used the databases to harass women. As cities cut off Flock’s funding, a death-spiral effect is emerging: with each city that leaves, the value of the shared national search network drops for the police departments that remain.
Germany Forces Apple to End Anticompetitive Self-Preferencing in Data Prompts · The Verge Germany’s Federal Cartel Office has forced Apple to redesign its App Tracking Transparency prompts after finding they were anticompetitive. Regulators accused Apple of self-preferencing by designing prompts that steer users away from consenting to data collection by third-party apps while giving its own apps a frictionless path to user data. Under the settlement, Apple must provide a level playing field, highlighting the increasing antitrust pressure Apple faces as a designated “gatekeeper” under the EU’s Digital Markets Act.
Also Worth Knowing
- [SpaceXAI Launches Grok Bot Cloud Agents for Multi-Step Workflows] (InfoQ): SpaceXAI has released Grok Bot, persistent AI agents running on dedicated cloud computers to perform end-to-end tasks across websites, databases, and inboxes.
- [Grab Automates Analytics Workflows to Cut Analyst Manual Load to 30%] (InfoQ): Grab’s Spartan system uses over 50 skills and 120 frameworks to automate data pulls and SQL requests, driving self-service rates up to 90%.
- [Cloudflare Converts CI Pipelines Into Durable TypeScript Workflows] (InfoQ): The new @cloudflare/ci SDK lets developers define continuous integration stages as TypeScript code that runs within checkpointed, resumable Cloudflare Workflows.
- [The Heartbreaking Planned Obsolescence of Moxie the Therapy Robot] (MIT Technology Review): The folding of Moxie’s new owner marks its second commercial failure, highlighting the emotional toll of planned obsolescence on neurodivergent children who form deep attachments to AI companions.
- [Supreme Court Rejects Verizon’s $47 Million FCC Refund Appeal] (Ars Technica): The high court denied Verizon’s bid to recoup a $47 million fine for selling customers’ real-time location data without consent, while AT&T and T-Mobile continue their battles.
🎧 I can convert this digest into a high-quality audio briefing if you’d like to listen to today’s tech stories on the go.