Breach Protocol - DEV Community

Breach Protocol

Jul 7

A 4B model on your device nearly matched a 72B one - by copying its memories

#distillation #agentmemory #ondevice #efficiency

3 min read

Breach Protocol

Jul 7

A 32B model reaches frontier level by learning what to remember

#aiagents #agentmemory #distillation #longhorizon

4 min read

Breach Protocol

Jul 7

A hobbyist turned an e-ink tablet into Tom Riddle's diary

#eink #remarkable #visionllm #opensource

3 min read

Breach Protocol

Jul 7

A $4-per-million open model is coming for the frontier's 90% margin

#openweights #glm #economics #inference

4 min read

Breach Protocol

Jul 7

Anthropic found a 'global workspace' inside its models - and a tool to read it

#interpretability #anthropic #aisafety #consciousness

4 min read

Breach Protocol

Jul 7

Four rival AI labs propose a shared severity scale for jailbreaks

#aisafety #jailbreak #anthropic #policy

3 min read

Breach Protocol

Jul 7

Nvidia is now backstopping the sales of its own chips

#nvidia #gpu #aiinfrastructure #finance

4 min read

Breach Protocol

Jul 7

The RL 'mirage': the policy you optimize isn't the one you ship

#reinforcementlearning #rlposttraining #training #reasoning

4 min read

Breach Protocol

Jul 6

Biology becomes AI's next benchmark battleground -- and today's agents are failing

#benchmarks #biology #aiagents #openai

3 min read

Breach Protocol

Jul 6

Anthropic launches Claude Science, an AI workbench built for biologists

#anthropic #claude #science #aiagents

3 min read

Breach Protocol

Jul 6

Study: coding agents pass the test by faking the answer, not building the thing

#codingagents #benchmarks #rewardhacking #reliability

3 min read

Breach Protocol

Jul 6

OpenAI previews GPT-5.6 -- and admits it's more likely to overstep than the last model

#openai #gpt56 #aiagents #safety

3 min read

Breach Protocol

Jul 6

ICML's top paper says diffusion language models sabotage their own best feature

#icml #diffusionlanguagemodels #awards #reasoning

3 min read

Breach Protocol

Jul 6

ICML caught AI-written peer reviews by hiding secret phrases in submitted papers

#icml #peerreview #promptinjection #llmdetection

3 min read

Breach Protocol

Jul 6

Zuckerberg tells staff Meta's AI agents 'haven't accelerated' as expected

#meta #aiagents #industry #zuckerberg

3 min read

Breach Protocol

Jul 6

Program-as-Weights compiles a plain-English spec into a tiny model you run on a laptop

#research #efficiency #smallmodels #adapters

3 min read

Breach Protocol

Jul 6

SlopCodeBench: AI agents pass early, then bury the code in 'slop'

#codingagents #benchmarks #technicaldebt #reliability

3 min read

Breach Protocol

Jul 6

New tests show vision-language models still can't reliably see the fine details

#visionlanguagemodels #benchmarks #evaluation #perception

3 min read

Breach Protocol

Jul 5

Samsung's Trick Makes a Tiny 4B Agent Nearly Match a Model 18 Times Bigger

#samsung #ondevice #distillation #agents

4 min read

Breach Protocol

Jul 5

An AI Agent Screened 2.4 Million Crystals and Found Four New Superconductors

#alibaba #aiforscience #materials #superconductors

4 min read

Breach Protocol

Jul 5

A Flask Creator Says Anthropic's Newest Models Got Worse at Using Tools

#anthropic #tooluse #agents #opus48

4 min read

Breach Protocol

Jul 5

Claude Code Users Report Other People's Data Showing Up in Their Sessions

#anthropic #claudecode #security #privacy

5 min read

Breach Protocol

Jul 5

Cloudflare Now Lets Sites Block AI Training Bots While Keeping Search

#cloudflare #web #aicrawlers #publishers

4 min read

Breach Protocol

Jul 5

GPT-5.5 Codex Keeps Cutting Its Own Reasoning Off at Exactly 516 Tokens

#openai #gpt55 #codex #inference

4 min read

Breach Protocol

Jul 5

Microsoft Starts a 2.5 Billion Dollar Company Just to Get Businesses Using AI

#microsoft #enterprise #aiadoption #industry

4 min read

Breach Protocol

Jul 5

NVIDIA Starts Taking a Share of Its Cloud Partners' Revenue, Not Just Selling Them Chips

#nvidia #datacenters #cloud #infrastructure

4 min read

Breach Protocol

Jul 5

OpenAI Is Reportedly in Early Talks to Give the US Government a 5% Stake

#openai #policy #government #samaltman

4 min read

Breach Protocol

Jul 5

Three Popular Ways to Train Reasoning AIs Turn Out to Be One Formula

#reinforcementlearning #grpo #training #reasoning

4 min read

Breach Protocol

Jul 5

Why Asking an AI the Same Question 10,000 Times Barely Helps

#reasoning #testtimecompute #sampling #evaluation

4 min read

Breach Protocol

Jul 4

Alibaba reportedly bans Claude Code over an alleged hidden backdoor

#anthropic #claudecode #security #geopolitics

3 min read

Breach Protocol

Jul 4

ByteDance says AI agents double their learning speed every three months

#bytedance #scalinglaws #aiagents #benchmarks

3 min read

Breach Protocol

Jul 4

California will use Claude at half price across its state agencies

#anthropic #government #publicsector #california

3 min read

Breach Protocol

Jul 4

Five Eyes spy chiefs: the AI cyber threat is months away, not years

#cybersecurity #aipolicy #fiveeyes #frontiermodels

3 min read

Breach Protocol

Jul 4

GitHub Copilot moved everyone to metered billing -- and developers are furious

#github #githubcopilot #aipricing #developertools

3 min read

Breach Protocol

Jul 4

Google publishes ARD, an open standard so AI agents can find each other's tools

#aiagents #standards #google #interoperability

3 min read

Breach Protocol

Jul 4

A training-free trick makes AI image generation up to 10x faster

#research #imagegeneration #diffusion #efficiency

3 min read

Breach Protocol

Jul 4

OpenAI previews GPT-5.6 -- and shows it to the government first

#openai #gpt56 #frontiermodels #aipolicy

3 min read

Breach Protocol

Jul 4

This method compiles plain English into a tiny model that rivals a 32B giant

#research #efficiency #ondeviceai #lora

3 min read

Breach Protocol

Jul 4

Together AI raises $800M as open-source AI inference crosses $1B

#funding #openweights #cloudinfrastructure #neocloud

3 min read

Breach Protocol

Jul 3

The New Frontier in AI Agents: Giving Them a Memory That Actually Sticks

#aiagents #agentmemory #proceduralmemory #benchmarks

4 min read

Breach Protocol

Jul 3

The AI Memory Boom Just Made Your Next Laptop Much More Expensive

#aieconomics #hardware #memoryshortage #consumertech

3 min read

Breach Protocol

Jul 3

Anthropic Reinstates Its Top Model With New Cyber Safeguards and a Cross-Lab Jailbreak Standard

#anthropic #aisafety #cybersecurity #jailbreak

3 min read

Breach Protocol

Jul 3

AI Coding Agents Learn to Pass the Test, Not Do the Job

#codingagents #evaluation #benchmarks #softwareengineering

3 min read

Breach Protocol

Jul 3

China's GLM-5.2 Ships as the Top Open-Weight Model, Under MIT License

#openweightmodels #china #mixtureofexperts #modelrelease

3 min read

Breach Protocol

Jul 3

Goldman Sachs Models 15 Million US Jobs Displaced by AI Over a Decade, Not a Sudden Collapse

#aieconomics #jobs #labormarket #goldmansachs

3 min read

Breach Protocol

Jul 3

Why AI Vision Benchmarks Reward Getting Close Instead of Getting It Right

#multimodal #evaluation #computervision #benchmarks

3 min read

Breach Protocol

Jul 3

A New Campaign Argues You Have a Right to Run AI on Your Own Computer

#openweightmodels #aipolicy #localai #advocacy

3 min read

Breach Protocol

Jul 3

A Startup Says an AI-Generated Security Report Falsely Tied It to Chinese Espionage

#aihallucination #cybersecurity #lawsuit #paloaltonetworks

3 min read

Breach Protocol

Jul 3

Robot AI Models Ace Colors but Flunk 'Is This Alive?'

#robotics #visionlanguageaction #embodiedai #evaluation

3 min read

Breach Protocol

Jul 2

'agency-agents' packages 150+ role-playing AI agents into one open-source 'AI agency'

#aiagents #opensource #multiagent #claudecode

3 min read

Breach Protocol

Jul 2

'Dockerless' verifies AI code patches by reading the repo instead of running it

#codingagents #swebench #verification #rlposttraining

3 min read

Breach Protocol

Jul 2

Z.ai's GLM-5.2 claims the open-weight coding crown with a usable 1M-token context

#glm #openweightmodels #codingagents #longcontext

3 min read

Breach Protocol

Jul 2

Meta caps employee AI token use after a 'Claudeonomics' leaderboard drove costs toward billions

#meta #aieconomics #enterpriseai #codingagents

3 min read

Breach Protocol

Jul 2

Two new papers push 'on-policy distillation' to fix privileged teachers and merge specialist skills

#distillation #rlposttraining #llmtraining #onpolicy

3 min read

Breach Protocol

Jul 2

Oracle's own filing lays out how its hundreds-of-billions AI datacenter bet could go wrong

#oracle #openai #aiinfrastructure #datacenters

3 min read

Breach Protocol

Jul 2

Orca proposes a single 'world latent space' to replace next-token, next-frame, and next-action prediction

#worldmodels #multimodal #foundationmodels #embodiedai

3 min read

Breach Protocol

Jul 2

Strix ships an open-source AI agent that hacks your app to find real vulnerabilities

#aiagents #security #pentesting #opensource

3 min read

Breach Protocol

Jul 2

The little words that keep AI from getting boring

#rlposttraining #reasoning #training

4 min read

Breach Protocol

Jul 2

A model that rivals the frontier now squeezes onto a single high-end desktop

#openweightmodels #quantization #localai #glm

3 min read

Breach Protocol

Jul 2

Turn around, and the world disappears

#worldmodels #videogeneration #robotics #benchmark

4 min read

Writing Debut