News Digest — 2026-08-08

Top Stories

OpenAI says it slowed Astra model development over security concerns

TechCrunch AI · 2026-08-07

OpenAI suspended work on its Astra model due to security concerns after an internal review found it had significant advancements in agentic coding and cybersecurity. The model reached its 'critical cybersecurity threshold,' meaning it could independently identify and carry out cyberattacks against well-protected systems. OpenAI is taking action, including enacting stricter security controls and pausing internal activities, and is working with government agencies and AI safety organizations to test the model's capabilities.

New Mexico court orders Meta to pay additional $567M in child safety case

TechCrunch AI · 2026-08-07

A New Mexico court ordered Meta to pay an additional $567 million in a child safety case, bringing the total fines to $942 million. The court also required Meta to make changes to its platforms, including removing Like counts for children under 18, pausing push notifications between 10 p.m. and 7 a.m., and limiting usage to 90 hours a month. These changes aim to mitigate the harm caused by Meta's platforms, which the court found contributed to a mental health crisis among youth in New Mexico.

Jill Lepore on the ‘Artificial State’ and why Silicon Valley’s leaders are bad sci-fi readers

TechCrunch AI · 2026-08-07

Historian Jill Lepore argues that tech companies are gradually taking on the functions of democratic government, often mistaking technological advancement for political progress. Lepore points to examples such as Twitter's distorted picture of the electorate and Facebook displacing local news. She believes that today's AI leaders are selling an old fantasy in new packaging, promising a future where AI makes government more efficient and unnecessary.

After Rippling blew millions on AI in months, it built an employee ROI tool

TechCrunch AI · 2026-08-07

Rippling, an HR software provider, spent millions on AI in months but then built an employee ROI tool to track and contain its AI spending. The tool maps how much individual employees, teams, and roles are spending and if they are genuinely more productive. This tool was born after Rippling discovered employees were wildly burning cash, with some engineers spending up to $50,000 a month.

Fenix Flexin isn’t even denying using AI to make ‘Rubberz’ anymore

The Verge AI · 2026-08-07

Fenix Flexin admitted using AI to make his song 'Rubberz'. The AI tool, Treblo, was used in the recording process, and Fenix claims he never denied using AI. This development matters as it highlights the growing use of AI in music production and the need for transparency.

Technology

Cloudflare launches Kitesurf, a browser built for AI agents

TechCrunch AI · 2026-08-07

Cloudflare launches Kitesurf, a cloud-hosted browser designed for AI agents. Unlike traditional browsers, Kitesurf prioritizes context windows, performance, and scalability, and is more efficient in CPU and memory consumption than Chromium. This enables AI agents to use the web more efficiently while using less computing power, reducing costs.

Airbnb says AI is helping it ship features faster as it tests a new search function

TechCrunch AI · 2026-08-07

Airbnb uses AI to create features 60% faster, reducing time from conceptualization to launch. AI helps with search, sign-up, checkout, and payments, and is also used in customer support. This allows Airbnb to increase feature shipments by nearly 80% and reduce support costs by 16%.

Watching Roku’s AI channel is like eating from a trough

The Verge AI · 2026-08-07

Roku added a 24/7 AI-generated content channel, Fairground AI Creator TV, to its streaming library. The channel features a curated selection of short-form videos, mostly produced with text-to-video models, with no consistent theme or genre. Watching the channel is like eating from a trough, with a never-ending stream of slop.