🥺 wikiHow sues OpenAI over AI training

wikiHow is suing OpenAI, alleging that the company scraped more than 11,000 of its articles to train GPT models without permission.

The complaint claims the copying infringed at least 1,200 registered copyrights, but the bigger argument goes beyond training.

wikiHow says ChatGPT can generate answers that effectively replace the original how-to articles, potentially hurting the site’s business.

OpenAI maintains that its models can use publicly available information under fair use, where courts also consider whether the use harms the market for the original work.

There’s an interesting legal wrinkle here, copyright protects the original expression of an article, but generally not the underlying facts, procedures or methods it describes. So ChatGPT explaining how to fix a sink isn’t automatically copyright infringement just because wikiHow also explained how to fix that sink.

The case was filed August 21 in the Southern District of New York and is still at the complaint stage. No court has ruled on the allegations.

Source.

@aipost 🏴

Labs are now putting people in motion suits to capture 'intimate' positions and movements in an effort to train humanoid dolls.

Was this on your 2026 bingo card?

@aipost 🏴

Qwen3.8-Flash-Next has been officially released, introducing a 6B-active open model that surpasses Claude Opus 4.6 Max on eight out of nine standard benchmarks.

The model applies a highly sparse Mixture of Experts (MoE) framework, featuring 125 billion overall parameters and 51 billion n-gram embeddings, with just 6 billion parameters active per token.

Benchmark results include:
• SWE-bench Pro: 62.5
• SWE-bench Multilingual: 81.0
• CoworkBench: 73.9
• JobBench: 55.7
• Toolathlon: 73.5
• IFBench: 81.3
• GPQA Diamond: 91.7
• LiveCodeBench: 91.9

Qwen3.8-Flash-Next also exceeds the performance of Qwen3.8-27B and DeepSeek-V4-Flash across the majority of evaluated categories.

📰 @aipost

🔥 Skild AI has introduced S1

A new robotics foundation model designed to learn new tasks from a single video demonstration without fine-tuning or post-training.

Skild says S1 can execute previously unseen tasks lasting up to 10 minutes, including:

• Making pour-over coffee
• Potting a plant
• Cooking pancakes
• Assembling a kit

The model takes a video demonstration as its prompt and translates it into robot actions for its own environment and embodiment.

Skild says S1 does not simply replay the demonstrated movements, it can recover from mistakes, handle environmental changes, substitute appropriate objects, and sometimes improve on errors made in the original demonstration.

@aipost 🏴

That pretty much sums up the AI race right now.

@aipost 🏴

❗️OpenAI’s executive exodus is getting hard to ignore

Another senior executive has left OpenAI.

Chris Malone, who led the company’s data center division, has departed after joining OpenAI in March 2025. He was responsible for a critical part of OpenAI’s massive infrastructure ambitions, including its plans to dramatically scale compute capacity.

And he’s far from the only one.

Since the start of 2026, OpenAI has seen a remarkable number of senior leaders leave, including executives overseeing commercial operations, research, robotics, Sora, marketing, security, risk, ethics, and infrastructure.

Among the departures:

• Head of Data Centers
• Chief Commercial Officer
• Chief Operating Officer
• Director of Communications
• Head of Robotics
• Director of Marketing
• Head of Research
• Head of Sora
• CTO of B2B
• CEO of AGI Deployment
• Head of Risk Assessment
• Chief Futurist
• Head of Security
• Lead AI Ethics Specialist

OpenAI is simultaneously restructuring its leadership, dramatically expanding its compute infrastructure, and preparing for a potential IPO.

The company says its infrastructure organization has been reorganized and that it has experienced leadership in place to execute its plans.

@aipost 🏴

AI will create jobs they said, starting here 😅

@aipost 🏴

Neuralink patient, Audrey is now able to create art with her mind.

@aipost 🏴