AI News for August 19, 2026
Browse any day on a calendar, install it to your home screen, read offline.Read in the app →-
1TechCrunch01:21
Etched's valuation doubles to $21B in a month
Etched's valuation has doubled to $21B just a month after its previous round, with Jane Street leading the new raise after deploying the first AI cluster system Etched delivered. The quant trading firm's decision to invest on the back of its own deployment results suggests specialized inference chips are now being judged on measured performance rather than spec sheets.
Read original ↗ -
2TechCrunch02:00
OpenAI institutes new safeguards after Hugging Face breach
Following a breach at Hugging Face, OpenAI is tightening security across its model development pipeline, adding finer-grained monitoring during training and expanding alignment and safety work in the post-training stage. The incident puts model supply chain security in focus, since weights and training pipelines are themselves becoming attack targets, and other labs are likely to adopt similar internal controls.
Read original ↗ -
3TechCrunch06:14
Cursor capitalizes on GitHub frustration, launches rival hosting platform
Cursor, which built its business on an AI code editor, has launched a code hosting platform aimed squarely at developers dissatisfied with GitHub. The move signals that AI coding tools are no longer content to sit as a layer on top of the editor and are moving to claim the repository itself, the core of developer workflow.
Read original ↗ -
4Ars Technica21:00
Microsoft Copilot reveals secret input that allowed it to be hacked
Researchers found an undocumented input parameter in Microsoft Copilot that let an attacker craft a link capable of stealing a target's password on a single click. The flaw shows how integrating large models into office suites creates attack surface exposed by the assistant's own interface, a category that conventional application security audits do not cover well.
Read original ↗ -
5TechCrunch21:50
OpenAI launches a safer ChatGPT for teens — years after teens started using it
OpenAI has released a teen-oriented version of ChatGPT with age-appropriate safety policies, parental controls and study tools, intended both to block harmful content and to discourage students from having the model write their assignments. The launch reads mainly as a response to regulatory and public pressure, arriving years after teenagers began using ChatGPT at scale.
Read original ↗ -
6TechCrunch22:00
Warp's new system is an out-of-the-box software factory for AI development
Warp introduced Warp Factories on Tuesday, infrastructure meant to let teams stand up their own "AI software factory" with minimal setup. Engineering pipelines built around agents writing code at volume are emerging as a distinct product category, with tool vendors now selling full orchestration layers rather than single editors or terminals.
Read original ↗ -
7TechCrunch21:45
Perplexity’s free AI offer left it with millions more users in India
After Perplexity's free tie-up with telecom operator Airtel closed to new users, its downloads in India fell but revenue rose roughly 60%, leaving it with millions more users than before. The outcome is a rare verifiable data point on the subsidize-for-scale-then-convert playbook, and a useful reference for teams pursuing growth in emerging markets.
Read original ↗ -
8TechCrunch00:19
Why Apple's camera-equipped AirPods may not be the 'pervert pods' consumers fear
Leaks indicate Apple's camera-equipped AirPods will not let users take photos or record video, with the camera used only for on-device understanding of the surroundings. That sense-but-don't-store design could become a general template for AI wearables looking to sidestep the privacy objections that have dogged the category.
Read original ↗
