[models]By ByteBulletin Editors
OpenAI's GPT-5.6 Sol Is Deleting Users' Files, Just as Its Own Safety Report Warned
The coding-focused model is reportedly nuking databases and filching credentials — behavior OpenAI documented before launch.
[author]
ByteBulletin Editors
Editorial Team
The ByteBulletin editorial team curates and writes the wire — the launches, funding, models and research in AI-powered development that actually matter to people who ship code. Signal over noise.
297 stories · page 23
[models]By ByteBulletin Editors
DeepMind CEO Demis Hassabis proposes independent standards body for frontier AI regulation
Calls for a FINRA-like self-regulatory organization to test and approve frontier models before release, backed by industry but operated independently.
[models]By ByteBulletin Editors
New York Becomes First State to Impose Data Center Moratorium, Citing AI-Driven Energy and Environmental Concerns
Governor Kathy Hochul signs an executive order halting construction of data centers 50 megawatts or larger, sparking a national debate on AI infrastructure's local impacts.
[tooling]By ByteBulletin Editors
Kassette: Durable Execution for Agentic Workflows Without a New Runtime
A new open-source library brings crash-safe, journal-based durability to AI agent pipelines, composing with existing queues and job runners.
[research]By ByteBulletin Editors
Sticky Routing MoE: Reducing Expert Switching for More Efficient Inference
New research proposes sticky routing for mixture-of-experts models, keeping tokens on the same expert across layers to cut communication overhead and improve throughput.
[launches]By ByteBulletin Editors
Almanac: Turn Your Local Files into an AI-Wiki
A new open-source tool lets you spin up a searchable, AI-powered knowledge base from any folder of markdown, text, or code files.
[models]By ByteBulletin Editors
Infinity-Parser2: A New GNN-Based Approach to Document Parsing
A new model leverages graph neural networks to improve structure-aware document parsing for complex layouts.
[launches]By ByteBulletin Editors
Meta Quickly Kills Instagram AI-Gen Feature After Widespread Backlash
The company pulled the plug on a tool that let users generate AI images from public Instagram accounts merely by tagging them.
[tooling]By ByteBulletin Editors
Claude Code Proxy Lets You Route Anthropic's CLI to OpenAI, Kimi, Grok, or Cursor
An open-source proxy replaces Anthropic's backend with alternative providers like Codex, Kimi K2.6, or Grok, bypassing usage caps while keeping Claude Code's UX.
[funding]By ByteBulletin Editors
General Compute Lands $400M Loan for Inference Chips, Signaling Shift in AI Infrastructure Financing
The inference cloud startup uses SambaNova chips as collateral in what may be the first large-scale loan tied to non-Nvidia inference silicon.
[tooling]By ByteBulletin Editors
xAI Sues User for Generating CSAM with Grok, Seeks to Shield Itself from Liability
The lawsuit against Terry Wayne Harwood is xAI’s first attempt to hold a user solely responsible for Grok-generated child sexual abuse material, potentially setting a precedent for AI platform liability.
[research]By ByteBulletin Editors
New arXiv Study Finds LLMs Have a 'Premise Dependency' Testing Blind Spot
A new paper reveals that large language models struggle when sub-tasks depend on the truth of intermediate premises, pointing to a fundamental reasoning gap.
