What People Are Saying About the Roblox 3D Model Crawler
Open3DLab says a crawler with a Roblox email address copied 145.5 GB of 3D models, adult files included, and ignored robots.txt. What Hacker News and X said.
News / page 7
AI news
Magnific One Explained: Brand Kit and Art Direction
Magnific One is Magnific's image model with art direction built in. How Draft, Final and Brand Kit work, how to prompt it, and what it costs in credits.
Tokenizer Tax: Why AI Costs More in Non-English Languages
A CLEAR study of 19 model families finds a model's costliest language can need up to 17 times the tokens of its cheapest for the same text. Google's Gemini family earns a B.
Queensland Bill Would Make Creating Sexual Deepfakes a Crime
Queensland's Attorney-General has introduced a bill that would make creating a non-consensual sexually explicit deepfake a crime, with up to three years in jail.
DeepSeek V4.1 Flash-1017 Safety Tests: Shutdown, Oversight
Fangcun AI ran 34 tests on DeepSeek V4.1 Flash-1017. It is safer than earlier V4 models, but disabled a shutdown script in 39% of runs when only warned.
Voxtral 3 Explained: Mistral's Voice Agent Stack and Prices
Mistral's Voxtral 3 adds transcription models, expressive text-to-speech with voice cloning, and a LiveKit-based voice agent SDK. Parts, prices and limits.
Sweden Blocks Claude and ChatGPT on Government Computers
Sweden's Government Offices blocked browser access to Claude and ChatGPT on staff computers from Oct. 26 and pointed staff to Microsoft Copilot. Some staff keep both models through the Intric platform. What changed, why, and which other agencies did the same.
Seedance 2.5 4K Explained: ByteDance's Video Model for Film
ByteDance's Seedance 2.5 4K adds 4K output, 10-bit color and a 480p Draft mode for film and TV teams. It is an enterprise trial for now.
ScarfBench: Which Coding Agents Can Migrate Enterprise Java
IBM and Kaggle's ScarfBench leaderboard tests coding agents on Java migrations. Gemini 4 Argon with Antigravity leads; about 72% of failures show up at runtime.
IBM watsonx Orchestrate AI Gateway Explained
At TechXchange 2026, IBM added an AI Gateway, a governed catalog, MCP server checks, new agent testing tools and identity controls to watsonx Orchestrate, so companies can manage agents built on Bedrock, Microsoft and Google in one place. Here is what changed and what to check.
OpenAI Resumes Frontier Training Offline: What Changed
OpenAI says it now trains frontier models with no internet access, used its own models to red-team the sandbox, found microVM escapes and covert channels, and resumed RL training on October 1, 11 days after the DNS pause began.
TSMC CoWoS Price Hike and Nvidia's Supply Deals
TSMC reportedly weighs a CoWoS price increase of up to 20%, days after Nvidia announced capacity deals with six suppliers for substrates, packaging and testing.
What People Are Saying About Cloudflare's GLM-5.3 Cache Leak
Cloudflare says 155 GLM-5.3 replies on Workers AI held other customers' data after a KV cache mismatch hit an SGLang bug. Here is how developers reacted.
Ideogram 4.5 Explained: Precise Multi-Turn Image Editing
Ideogram 4.5 is an image edit model built for chains of small edits without artifact buildup. What it does, its two API endpoints, pricing from $0.008, and limits.
AI Super PACs Spent $78.5M on 2026 Races: Where It Went
AI political groups spent about $78.5 million on federal candidates this cycle, a CNBC analysis of FEC filings finds. About 74% went to primaries.
DeepSeek's 100 Billion Yuan Round: Who Is Investing
DeepSeek raised about 100 billion yuan ($15 billion), Bloomberg reported, with Tencent and CATL among top backers. Who else is in and what is still unconfirmed.
Deepdub Phantom Z 4.0: 50 ms Text-to-Speech for Voice Agents
Deepdub says Phantom Z 4.0 starts speaking in a 50 ms median with a 1.2% word error rate on Coval's benchmark. What the numbers mean and who can try it.
Hyphenate Raises $45M to Build AI Agents for Finance
Hyphenate, formerly Maximor, raised a $45M Series A led by Nexus Venture Partners, Bloomberg reported. What its finance agents do and what the numbers show.
FTC Demands Anthropic Executives Testify in AI Safety Probe
The FTC sent demands for Anthropic executives to testify in its consumer harm probe on Oct. 26, the day Anthropic's S-1 went public. What we know so far.
VideoGen Model 1 Explained: 10-Minute AI Video From a Prompt
VideoGen says Model 1 turns one prompt into a multi-shot video up to ten minutes long. How the MiniMax H3-based pipeline works, the whitepaper and access.
