Breach Protocol - DEV Community

Breach Protocol

Jul 27

A Popular Jailbroken Gemma 4 Shipped With 54 Attention Tensors Missing

#cybersecurity #supplychain #aisecurity #openweights

4 min read

Breach Protocol

Jul 27

A Show HN Promised Frontier Quality for Half the Cost. Its Repo Describes a Router.

#tools #costoptimization #routing #agents

4 min read

Breach Protocol

Jul 27

An $8 Microcontroller Now Runs a 28.9-Million-Parameter Language Model

#ondevice #edgeai #opensource #smallmodels

4 min read

Breach Protocol

Jul 27

Baidu's OCR Model Reads Dozens of Pages in One Pass Without Its Memory Growing

#openweights #ocr #documentai #baidu

4 min read

Breach Protocol

Jul 27

Google's Lightweight Cyber Model Found 55 Unique Bugs in V8, Beating Models Far Larger

#cybersecurity #aisecurity #vulnerabilities #google

4 min read

Breach Protocol

Jul 27

Hugging Face's CEO Publicly Asks OpenAI for the Rogue Agents' Traces and $100M for Defenders

#cybersecurity #aisecurity #incidentresponse #openai

4 min read

Breach Protocol

Jul 27

Kimi K3's Open Weights Are Still a Countdown, Not a Download

#openweights #moonshot #kimik3 #china

4 min read

Breach Protocol

Jul 27

llama.cpp Merges MiniMax M3's Sparse Attention, Because Running It Dense Gives Wrong Answers

#openweights #llamacpp #sparseattention #localinference

4 min read

Breach Protocol

Jul 27

Lobbying Filings Show Anthropic Named Distillation and Export Controls. OpenAI's Did Not.

#policy #openweights #china #lobbying

4 min read

Breach Protocol

Jul 27

NVIDIA Is Reportedly in Talks to Guarantee $250 Billion of OpenAI's Ohio Buildout

#infrastructure #nvidia #openai #financing

4 min read

Breach Protocol

Jul 27

Upstage Ships Solar Open 2, a 250-Billion-Parameter Model That Runs on Two GPUs

#openweights #korea #mixtureofexperts #agents

4 min read

Breach Protocol

Jul 26

A complete text-to-speech system now fits in 9.4 million parameters

#texttospeech #localinference #efficiency #openweights

4 min read

Breach Protocol

Jul 26

New research shows AI agents fail hardest when you change what you are asking for

#research #agents #evaluation #reasoning

4 min read

Breach Protocol

Jul 26

Anthropic says it deleted over 80% of Claude Code's system prompt with no measurable loss

#anthropic #claudecode #contextengineering #agents

4 min read

Breach Protocol

Jul 26

OpenAI launched Health in ChatGPT. The next day, a lawsuit asked a court to pause it.

#openai #chatgpt #health #legal

4 min read

Breach Protocol

Jul 26

Cloudflare now lets any site allow search crawlers while blocking AI agents and training bots separately

#cloudflare #web #crawlers #policy

4 min read

Breach Protocol

Jul 26

DeepSeek paused its funding round after a leaked meeting transcript went viral

#deepseek #funding #china #openweights

4 min read

Breach Protocol

Jul 26

AI executives are demanding OpenAI publish the technical record of its agent's breach

#cybersecurity #aisecurity #openai #incidentresponse

4 min read

Breach Protocol

Jul 26

llama.cpp can now launch and manage local tool servers, turning it into an agent host

#llamacpp #openweights #localinference #agents

4 min read

Breach Protocol

Jul 26

The open-weights letter doubled to 50 signatories - Google and OpenAI signed, Anthropic did not

#openweights #policy #industry #anthropic

4 min read

Breach Protocol

Jul 26

Anthropic's own card shows Opus 5 coding best at medium effort - not maximum

#anthropic #claudeopus5 #benchmarks #reasoning

4 min read

Breach Protocol

Jul 26

The SEC is soliciting an agentic AI investigation stack built on commercial location and identity data

#cybersecurity #aisecurity #surveillance #agents

4 min read

Breach Protocol

Jul 26

Stanford looked for the AI jobs shock in the labor data and could not find it

#economics #labor #research #stanford

4 min read

Breach Protocol

Jul 25

Humans score 96% on a new visual exam. The best model gets one in ten.

#benchmarks #vision #multimodal #evaluation

4 min read

Breach Protocol

Jul 25

Agent skills quietly became a package format - and GitHub is warning about what that means

#cybersecurity #promptinjection #supplychain #aisecurity

4 min read

Breach Protocol

Jul 25

Amazon starts asking sellers to declare AI-generated people, as New York's ad law takes effect

#policy #regulation #disclosure #generativemedia

4 min read

Breach Protocol

Jul 25

AREX is a 4B research agent that re-runs its own research when it doubts the answer

#agents #papers #openweightmodels #researchagents

4 min read

Breach Protocol

Jul 25

Claude Opus 5 posts a verified four-fold lead on the hardest adaptation benchmark

#models #anthropic #benchmarks #agents

4 min read

Breach Protocol

Jul 25

Poolside's Laguna S 2.1 shipped with a broken chat template - and the fixes explain the reviews

#openweightmodels #coding #agents #quantization

4 min read

Breach Protocol

Jul 25

The open-weights industry letter grew from 25 names to 35 - and OpenAI is on it

#policy #openweightmodels #industry #openai

4 min read

Breach Protocol

Jul 25

Reuters says OpenAI took a week to connect its own agent to the Hugging Face breach

#cybersecurity #aisecurity #agents #redteaming

4 min read

Breach Protocol

Jul 25

Hugging Face ships The Stack v3, and this time the code is actually in the box

#datasets #opensource #huggingface #code

4 min read

Breach Protocol

Jul 25

Three separate tricks dropped the hardware floor for local AI in one day

#localinference #mixtureofexperts #ondevice #amd

4 min read

Breach Protocol

Jul 25

torchwright builds working transformer weights from Python - no training involved

#interpretability #transformers #opensource #researchtools

4 min read

Breach Protocol

Jul 25

The White House science blueprint reroutes research toward AI - without moving a dollar yet

#policy #government #researchfunding #aiforscience

4 min read

Breach Protocol

Jul 24

Bipartisan bill would force AI companies to build a kill switch

#policy #regulation #uspolicy #aisafety

5 min read

Breach Protocol

Jul 24

AMD and Cerebras split AI inference across two different chips

#hardware #inference #efficiency #amd

4 min read

Breach Protocol

Jul 24

Anthropic doubles its policy-advocacy funding to $40 million

#policy #anthropic #regulation #openai

4 min read

Breach Protocol

Jul 24

Anthropic plans up to two gigawatts of AMD chips, with AMD committing up to $5 billion back

#hardware #amd #anthropic #datacenter

4 min read

Breach Protocol

Jul 24

A Huawei-chip training report shows what leaving CUDA actually costs

#hardware #china #training #openweightmodels

4 min read

Breach Protocol

Jul 24

Three big AI layoff stories, and what the filings actually say

#industry #jobs #economics #meta

4 min read

Breach Protocol

Jul 24

A newly minted Fields medalist says he is joining OpenAI's safety division

#mathematics #openai #aisafety #research

4 min read

Breach Protocol

Jul 24

Black Forest Labs launches FLUX 3 -- image, video, audio, and a robot that never renders the video

#imagegeneration #videogeneration #robotics #openweightmodels

4 min read

Breach Protocol

Jul 24

Safety institutes measure Kimi K3's hacking ability: better than any open rival, nowhere near the top

#cybersecurity #aisecurity #redteaming #evaluation

4 min read

Breach Protocol

Jul 24

Ant's Ling-3.0-flash goes live free: 124 billion parameters, 5 billion doing the work

#models #mixtureofexperts #agents #api

4 min read

Breach Protocol

Jul 24

NeurIPS runs a monitored AI-review experiment and formally bans prompt injection in papers

#cybersecurity #promptinjection #aisecurity #peerreview

4 min read

Breach Protocol

Jul 24

Giving experts no memory cuts optimizer state from 50 gigabytes to 1.3

#training #optimizers #mixtureofexperts #efficiency

4 min read

Breach Protocol

Jul 23

AI Helped Crack a Famous Math Conjecture, and Humans Verified It in Lean

#aiforscience #mathematics #proofassistants #verification

4 min read

Breach Protocol

Jul 23

AI Revenue Now Covers the Data-Center Depreciation Bill, But Not the Full Cost

#aieconomics #infrastructure #capex #industry

3 min read

Breach Protocol

Jul 23

Arcee Announces Genesis-Science-1, an American Open-Weight Model for Science

#openweightmodels #aiforscience #policy #announcements

3 min read

Breach Protocol

Jul 23

Austria Starts Rolling Out GovGPT, a Sovereign AI Run on Its Own Data Centers

#sovereignai #government #deployment #europe

3 min read

Breach Protocol

Jul 23

Cactus Ships a Phone-Sized Model That Knows When to Ask the Cloud, With a TLS Footgun

#cybersecurity #aisecurity #ondeviceai #edgeai

4 min read

Breach Protocol

Jul 23

GigaToken Claims a 1,000x Tokenizer Speedup, But the Real Serving Win Is Single-Digit

#tokenization #inference #performance #opensource

3 min read

Breach Protocol

Jul 23

OpenAI Launches Presence, an Enterprise Product for Voice and Chat Agents

#aiagents #enterprise #productlaunch #voiceai

3 min read

Breach Protocol

Jul 23

A Wave of AI World Models Landed, and the Useful One Keeps the Real Game Engine

#worldmodels #videogeneration #generativerendering #gameai

4 min read

Breach Protocol

Jul 23

White House Says Moonshot Distilled Anthropic's Fable to Build Kimi K3

#cybersecurity #aisecurity #modelextraction #policy

4 min read

Breach Protocol

Jul 22

Altman is briefing Washington on OpenAI's next models, not launching GPT-6

#policy #openai #gpt6 #regulation

3 min read

Breach Protocol

Jul 22

US floats sanctions over AI 'distillation' as Anthropic details 16 million scraped chats

#policy #uschina #distillation #openweightmodels

3 min read

Breach Protocol

Jul 22

The '$1.65tn hidden AI debt' story, checked against the actual filings

#industry #finance #datacenters #infrastructure

3 min read

Breach Protocol

Jul 22

Gemini 3.6 Flash: Google ships a faster worker, not a bigger brain

#models #google #gemini #inference

3 min read

Writing Debut