📱 Telegram adds built-in AI text editor with Cocoon models

Telegram rolled out an AI editor that appears after typing a few words. The “Ai” button lets users translate text, fix grammar, or change style. It runs on open source models via Cocoon Network.

The feature sits directly in the input field and works as a quick rewrite tool inside chats. No separate app or workflow.

The update also brings media attachments to polls, Live Photos support on iOS, a bot builder, a document scanner on iOS, plus labels for unofficial clients.

📊 @tech

🚕 100+ Baidu robotaxis stop mid-ride after system failure in China

More than 100 Baidu robotaxis suddenly halted during trips. Some stopped on busy roads, leaving passengers unable to exit safely.

The vehicles run as part of Baidu’s autonomous ride service. The incident shows how a single failure can affect many cars at once.

A similar case happened in December when Waymo vehicles stalled during a major power outage in San Francisco.

📊 @tech

🚀 NASA launches Artemis II toward the Moon after 54-year gap

NASA’s Space Launch System has launched the Artemis II mission to the Moon.

It marks the first crewed lunar flight in 54 years.

More than 3 million people watched the launch live.

First images from the mission are expected in 6 days.

📊 @tech

🤖 Claude has announced that computer use in Claude CoWork and Claude Code Desktop is now available on Windows.

📊 @tech

🤖 Anthropic targets 8,100 GitHub repos over Claude Code leaks

Anthropic sent DMCA requests against around 8,100 repositories hosting copies or forks of Claude Code. GitHub removed most of them within hours.

Some takedowns hit modified or unrelated projects. After уточнение требований, only 96 repositories remain under action, while the rest were restored. Projects like openclaude are still accessible.

Claude Code is Anthropic’s developer tooling tied to its models. The company is now actively enforcing control over its distribution.

📊 @tech

📱 Apple opens CarPlay to AI apps as ChatGPT launches on iOS 26.4

Apple added a new app category in CarPlay with iOS 26.4, allowing voice AI assistants. On March 31, OpenAI released ChatGPT for CarPlay as the first major app.

The interface is fully voice-based. No text or visuals on screen. Users open the app from the car display, speak, and hear responses. It mirrors the iPhone voice mode but adapted for driving. Activation is manual via the icon.

Support for other assistants like Claude and Gemini was reported earlier in February.

📊 @tech

📱 Apple bans vibe-coding apps as developers move them into iMessage

Apple removed apps like Replit, Vibecode, and Anything from the App Store, citing rules against executing code that changes app behavior after installation.

These tools allowed users to generate and run apps directly on iPhone without traditional development, which conflicts with App Store policies.

Developers behind Anything pushed an update that moves the functionality into iMessage, letting users build iOS apps without installing the main app.

Restrictions at the store level are pushing dev tools into alternative entry points.

📊 @tech

🌍 NASA releases Orion photos of Earth with auroras and zodiacal light

NASA published the first images from Orion during Artemis II. Commander Reid Wiseman captured Earth through the spacecraft window after the burn toward the Moon.

The photo shows two auroras and zodiacal light while Earth blocks the Sun.

It was shot on a Nikon D5 with f/4.0, 1/4s, 22 mm, ISO 51200. Resolution is 5568 × 3712, just over 20 MP.

📊 @tech

🤖 Anthropic cuts Claude subscriptions from third-party tools like OpenClaw

Since April 4, Anthropic disabled using Claude subscriptions inside external services. OpenClaw still works with Claude accounts, but usage is now billed separately via API token pricing.

The change targets heavy usage cases like large contexts and agent workflows, which exceeded the fixed $20 subscription model.

Workarounds rely on changing how requests are labeled so they look like first-party usage. Tool names get rewritten to pass filters.

📊 @tech

📱 Netflix releases VOID, an open model for removing objects from video

Netflix published VOID, a model that removes objects from video while rebuilding scene physics. If a person holding a guitar is removed, the guitar falls naturally instead of staying frozen.

The system combines components from CogVideoX by Alibaba, Gemini 3 Pro by Google, SAM2 by Meta, plus data from Adobe. In blind tests, VOID was preferred in 64.8% of cases, while Runway scored 18.4%.

Available on Hugging Face: https://huggingface.co/netflix/void-model

📊 @tech