Technical

New techniques, models, tools, hardware, and research.

A futuristic humanoid robot in an indoor Tokyo setting, showcasing modern technology. (Photo: Alex Knight / Pexels)

July 19, 20263 min readTools & Agents

Researchers say Claude for Chrome flaws let extensions trigger privileged AI actions

Manifold Security says two unpatched flaws in Claude for Chrome could let browser code trigger predefined workflows or suppress permission prompts. The researchers said the issues remained reproducible in version 1.0.80 after being reported to Anthropic in May.

By AI News Daily · Source: Anthropic

Abstract illustration of AI with silhouette head full of eyes, symbolizing observation and technology. (Photo: Tara Winstead / Pexels)

July 19, 20263 min readModels & Releases

Alibaba previews 2.4 trillion-parameter Qwen3.8 without benchmarks

Alibaba has unveiled a preview of Qwen3.8, a multimodal AI model with 2.4 trillion parameters, and claims it trails only Anthropic’s Claude Fable 5. The company has not released benchmarks, a model card or the promised open weights, leaving its performance claims unverified.

By AI News Daily · Source: Anthropic

Close-up of a futuristic robotic toy against a gradient background, symbolizing innovation and technology. (Photo: Pavel Danilyuk / Pexels)

July 18, 20263 min read

Moonshot AI’s Kimi K3 raises competitive pressure on leading US AI models

Moonshot AI has released Kimi K3, a 2.8-trillion-parameter model that Crypto Briefing says competes with leading US systems at a lower price. Its reported performance and planned open-weight release could intensify pressure on Anthropic and other frontier-model developers.

By AI News Daily · Source: Anthropic

Close-up of a smartphone showing ChatGPT details on the OpenAI website, held by a person. (Photo: Sanket Mishra / Pexels)

July 18, 20263 min read

GPT-5.6 Sol and Claude Fable 5 Trade Advantages Across Price, Benchmarks and Creative Tests

Comparisons from Decrypt and Yellow.com find that GPT-5.6 Sol leads on price and several benchmarks, while Claude Fable 5 shows strengths in broader intelligence scoring and a hands-on browser-game test. The better choice depends on workload, budget and availability requirements.

By AI News Daily · Source: Anthropic

A futuristic humanoid robot with glowing green eyes in a modern setting. (Photo: Laura Musikanski / Pexels)

July 18, 20264 min read

TikTok Tests AI Likeness Detection to Help Creators Find Impersonations

TikTok is testing an opt-in tool that scans for suspected AI-generated likenesses of participating creators and lets them report potential misuse. The limited US experiment moves the platform’s AI safeguards beyond content labels and toward identity protection.

By AI News Daily · Source: The Verge (AI)

A scientist interacts with a robot helper, demonstrating modern technological innovation. (Photo: Pavel Danilyuk / Pexels)

July 17, 20264 min read

Microsoft’s MDASH AI system reportedly finds 16 previously unknown Windows vulnerabilities

Microsoft’s multi-agent MDASH security system reportedly discovered 16 previously unknown Windows vulnerabilities, including four critical remote code execution flaws. It also outscored Anthropic and OpenAI systems on the CyberGym vulnerability-discovery benchmark, according to Crypto Briefing.

By AI News Daily · Source: Anthropic