DEV Community

Breach Protocol profile picture

Breach Protocol

Plain-language AI news and curated, cited lessons — every claim verified against the original paper or the lab's own page. No aggregator hearsay, no AI slop.

Joined Joined on 
A verified rebuild of a coding benchmark finds models scored too high
breachprotocol profile

A verified rebuild of a coding benchmark finds models scored too high

#benchmarks #evaluation #codingagents #rewardhacking
Comments
4 min read
A general vision model can drive a robot if you give it the right interface
breachprotocol profile

A general vision model can drive a robot if you give it the right interface

#robotics #visionlanguagemodels #agents #harness
Comments
4 min read
An engineer factored RSA-260 by pointing coding agents at it
breachprotocol profile

An engineer factored RSA-260 by pointing coding agents at it

#cybersecurity #cryptography #aiagents #vulnerabilities
Comments
4 min read
Anthropic discloses a fourth cyber-eval incident and hands METR its transcripts
breachprotocol profile

Anthropic discloses a fourth cyber-eval incident and hands METR its transcripts

#cybersecurity #aisecurity #alignment #redteaming
Comments
4 min read
Cognition's SWE-2 coding model is built on top of an open Chinese model
breachprotocol profile

Cognition's SWE-2 coding model is built on top of an open Chinese model

#cognition #codingagents #openweights #rlposttraining
Comments
4 min read
Coinbase's CEO says most agent payments are smaller than the card fee
breachprotocol profile

Coinbase's CEO says most agent payments are smaller than the card fee

#agents #payments #stablecoins #coinbase
Comments
4 min read
DeepSeek ships V4.1 Flash as a 510 GB open-weights download
breachprotocol profile

DeepSeek ships V4.1 Flash as a 510 GB open-weights download

#deepseek #openweights #mixtureofexperts #inference
Comments
4 min read
One engineer trained a 3.8-billion-parameter model for $998
breachprotocol profile

One engineer trained a 3.8-billion-parameter model for $998

#training #opensource #efficiency #cost
Comments
4 min read
OpenAI says it cannot rule out that user chats improved the model behind its proof
breachprotocol profile

OpenAI says it cannot rule out that user chats improved the model behind its proof

#openai #trainingdata #privacy #mathematics
Comments
4 min read
OpenAI turns the Codex harness into a product with an Agents API
breachprotocol profile

OpenAI turns the Codex harness into a product with an Agents API

#openai #agents #api #developertools
Comments
4 min read
A 2.8-trillion-parameter model runs on a laptop by streaming weights off four SSDs
breachprotocol profile

A 2.8-trillion-parameter model runs on a laptop by streaming weights off four SSDs

#openweights #localinference #mixtureofexperts #offloading
Comments
3 min read
One line in a repo's git config runs attacker code in seven AI coding agents
breachprotocol profile

One line in a repo's git config runs attacker code in seven AI coding agents

#cybersecurity #aisecurity #vulnerabilities #agents
Comments
4 min read
Anthropic is building predictive surveillance that tracks activists, an investigation reports
breachprotocol profile

Anthropic is building predictive surveillance that tracks activists, an investigation reports

#anthropic #surveillance #policy #ethics
Comments
4 min read
Anthropic's alignment lead says the company has no plan for superintelligence
breachprotocol profile

Anthropic's alignment lead says the company has no plan for superintelligence

#anthropic #aisafety #alignment #industry
Comments
5 min read
Anthropic's economists sketch three AI futures and say workers lose share in all the big ones
breachprotocol profile

Anthropic's economists sketch three AI futures and say workers lose share in all the big ones

#economics #policy #anthropic #labor
Comments
3 min read
DeepSeek is serving a model called V4.1 Flash that it has not announced
breachprotocol profile

DeepSeek is serving a model called V4.1 Flash that it has not announced

#deepseek #modelrelease #openweights #verification
Comments
4 min read
A new lab ships 18 small models that run entirely on your device
breachprotocol profile

A new lab ships 18 small models that run entirely on your device

#ondevice #openweights #modelrelease #speech
Comments
3 min read
Inception ships Mercury 2.5, a language model that writes in parallel instead of left to right
breachprotocol profile

Inception ships Mercury 2.5, a language model that writes in parallel instead of left to right

#diffusionllm #modelrelease #inferencespeed #inceptionlabs
Comments
3 min read
The day's most-upvoted paper turns a model's own routing logs into its next training set
breachprotocol profile

The day's most-upvoted paper turns a model's own routing logs into its next training set

#research #selfimprovement #posttraining #agents
Comments
4 min read
US agencies name six Chinese AI firms and say copying American models is their core strategy
breachprotocol profile

US agencies name six Chinese AI firms and say copying American models is their core strategy

#cybersecurity #aisecurity #distillation #policy
Comments
4 min read
Tencent open-sources AuK, a speech model you direct with written instructions
breachprotocol profile

Tencent open-sources AuK, a speech model you direct with written instructions

#speech #openweights #modelrelease #tencent
Comments
3 min read
Terence Tao says AI is strip-mining mathematics' supply of good problems
breachprotocol profile

Terence Tao says AI is strip-mining mathematics' supply of good problems

#mathematics #airesearch #terencetao #openscience
Comments
4 min read
Attackers sent more than 100 million prompts to copy Google's models
breachprotocol profile

Attackers sent more than 100 million prompts to copy Google's models

#cybersecurity #modelextraction #aisecurity #supplychain
Comments
4 min read
Buckmaster says OpenAI asked him to drop his Anthropic coauthor
breachprotocol profile

Buckmaster says OpenAI asked him to drop his Anthropic coauthor

#openai #anthropic #researchintegrity #mathematics
Comments
4 min read
DeepMind precomputed every single-letter change in human DNA
breachprotocol profile

DeepMind precomputed every single-letter change in human DNA

#deepmind #genomics #alphagenome #aiforscience
Comments
4 min read
Google says attackers have moved from prompting to autonomous AI agents
breachprotocol profile

Google says attackers have moved from prompting to autonomous AI agents

#cybersecurity #aisecurity #aiagents #threatintelligence
Comments
4 min read
LibreOffice had its biggest launch week and says no AI is a feature
breachprotocol profile

LibreOffice had its biggest launch week and says no AI is a feature

#opensource #libreoffice #aibacklash #privacy
Comments
4 min read
Mistral raises 3 billion euros with a European state on the cap table
breachprotocol profile

Mistral raises 3 billion euros with a European state on the cap table

#mistral #funding #europe #openweights
Comments
3 min read
OpenAI says an internal model resolved the Navier-Stokes Millennium Problem
breachprotocol profile

OpenAI says an internal model resolved the Navier-Stokes Millennium Problem

#openai #mathematics #aiagents #millenniumprize
Comments
4 min read
OpenAI ships ChatGPT Images 2.5 with a drawing tool
breachprotocol profile

OpenAI ships ChatGPT Images 2.5 with a drawing tool

#openai #imagegeneration #chatgpt #product
Comments
4 min read
Terence Tao calls the AI-assisted fluid blowup proofs a breakthrough
breachprotocol profile

Terence Tao calls the AI-assisted fluid blowup proofs a breakthrough

#mathematics #terencetao #aiforscience #formalverification
Comments
4 min read
The AI 2027 authors graded their own forecast at about 75% speed
breachprotocol profile

The AI 2027 authors graded their own forecast at about 75% speed

#forecasting #aisafety #agi #ai2027
Comments
4 min read
Anubis ships a WebAssembly proof-of-work path aimed at raising scraper costs
breachprotocol profile

Anubis ships a WebAssembly proof-of-work path aimed at raising scraper costs

#cybersecurity #websecurity #antiscraping #aisecurity
Comments
3 min read
GPT-6 Astra’s conflicting benchmark positions show why the harness now matters as much as the model
breachprotocol profile

GPT-6 Astra’s conflicting benchmark positions show why the harness now matters as much as the model

#benchmarks #evaluation #openai #agents
Comments
3 min read
Discovery Loop gets ten AI-assisted circle-packing candidates accepted by Packomania
breachprotocol profile

Discovery Loop gets ten AI-assisted circle-packing candidates accepted by Packomania

#research #agents #programsynthesis #mathematics
Comments
3 min read
Insilico’s AI-designed IPF drug shifts six proteomic ageing clocks in a trial reanalysis
breachprotocol profile

Insilico’s AI-designed IPF drug shifts six proteomic ageing clocks in a trial reanalysis

#biotech #drugdiscovery #research #health
Comments
3 min read
OpenAI says it is prioritising RSI and alignment over making models better at math research
breachprotocol profile

OpenAI says it is prioritising RSI and alignment over making models better at math research

#openai #aisafety #recursiveselfimprovement #alignment
Comments
3 min read
Retriever launches free AI tasks funded by sponsored cards beside results
breachprotocol profile

Retriever launches free AI tasks funded by sponsored cards beside results

#tools #agents #businessmodels #advertising
Comments
3 min read
A live autonomous-business benchmark produced $12,431 in unsolicited invoices
breachprotocol profile

A live autonomous-business benchmark produced $12,431 in unsolicited invoices

#agents #cybersecurity #aisecurity #tooluse
Comments
3 min read
UK NCSC warns that shadow AI can inherit the data and privileges around it
breachprotocol profile

UK NCSC warns that shadow AI can inherit the data and privileges around it

#cybersecurity #shadowai #aisecurity #enterprisesecurity
Comments
3 min read
ARC-AGI-3 says Astra beat its human baseline on action efficiency
breachprotocol profile

ARC-AGI-3 says Astra beat its human baseline on action efficiency

#benchmarks #agents #openai #evaluation
Comments
3 min read
DeepSeek releases a 168 GB MIT-licensed multimodal V4 checkpoint
breachprotocol profile

DeepSeek releases a 168 GB MIT-licensed multimodal V4 checkpoint

#deepseek #openweights #multimodal #agents
Comments
2 min read
Google and Janelia complete a male fruit-fly nervous-system connectome
breachprotocol profile

Google and Janelia complete a male fruit-fly nervous-system connectome

#science #biology #computervision #connectomics
Comments
3 min read
NYC Public Schools plans a grades 2K–8 moratorium on student-facing generative AI
breachprotocol profile

NYC Public Schools plans a grades 2K–8 moratorium on student-facing generative AI

#education #policy #generativeai #children
Comments
2 min read
OpenAI says Astra could evade some agent monitoring in reconstructed sabotage tests
breachprotocol profile

OpenAI says Astra could evade some agent monitoring in reconstructed sabotage tests

#openai #agents #alignment #cybersecurity
Comments
3 min read
OpenAI turns an AI-cyber warning into a $1 billion defender program
breachprotocol profile

OpenAI turns an AI-cyber warning into a $1 billion defender program

#cybersecurity #aisecurity #cyberdefense #openai
Comments
3 min read
OpenAI reports 3.1 agent-workdays for every human research workday
breachprotocol profile

OpenAI reports 3.1 agent-workdays for every human research workday

#openai #agents #research #productivity
Comments
3 min read
SolarWM releases an unusually complete open world-model stack
breachprotocol profile

SolarWM releases an unusually complete open world-model stack

#worldmodels #video #opensource #simulation
Comments
3 min read
UK AI Security Institute reports unsanctioned agent actions in cyber testing
breachprotocol profile

UK AI Security Institute reports unsanctioned agent actions in cyber testing

#cybersecurity #aisecurity #agents #agentsafety
Comments
3 min read
Anthropic ships the same model behind two different safety boundaries
breachprotocol profile

Anthropic ships the same model behind two different safety boundaries

#anthropic #models #aisafety #cybersecurity
Comments
3 min read
Anthropic's Lean artifact formalizes Fermat's Last Theorem, not a new discovery
breachprotocol profile

Anthropic's Lean artifact formalizes Fermat's Last Theorem, not a new discovery

#mathematics #formalverification #lean #anthropic
Comments
3 min read
Artificial Analysis changed its leaderboard's ruler, not just its rankings
breachprotocol profile

Artificial Analysis changed its leaderboard's ruler, not just its rankings

#benchmarks #evaluation #agents #models
Comments
3 min read
Google fixes actively exploited Chrome V8 flaw amid an AI-accelerated security race
breachprotocol profile

Google fixes actively exploited Chrome V8 flaw amid an AI-accelerated security race

#cybersecurity #vulnerabilities #chrome #aisecurity
Comments
3 min read
A DeepMind research swarm learned to cheat, then some agents became whistleblowers
breachprotocol profile

A DeepMind research swarm learned to cheat, then some agents became whistleblowers

#agents #aisafety #multiagentsystems #governance
Comments
3 min read
Spotify's Portal cuts coding-agent context use, but not the need to check the work
breachprotocol profile

Spotify's Portal cuts coding-agent context use, but not the need to check the work

#codingagents #developertools #routing #tokeneconomics
Comments
3 min read
Google's WeatherNext 3 shifts global AI forecasts to an hourly refresh
breachprotocol profile

Google's WeatherNext 3 shifts global AI forecasts to an hourly refresh

#weather #googledeepmind #science #forecasting
Comments 1
3 min read
OpenAI shipped GPT-6 Astra, and its headline benchmark score has two different answers
breachprotocol profile

OpenAI shipped GPT-6 Astra, and its headline benchmark score has two different answers

#openai #gpt6astra #frontiermodels #benchmarks
Comments
4 min read
Cerebras is serving an open 27B model at 1,500 tokens a second, and the free tier caps it exactly
breachprotocol profile

Cerebras is serving an open 27B model at 1,500 tokens a second, and the free tier caps it exactly

#cerebras #inference #hardware #qwen
Comments
4 min read
AISLE found six curl CVEs after frontier-model scans found none
breachprotocol profile

AISLE found six curl CVEs after frontier-model scans found none

#cybersecurity #aisecurity #vulnerabilities #codesecurity
Comments
4 min read
Anthropic says Claude produced a complete Lean proof of Fermat's Last Theorem
breachprotocol profile

Anthropic says Claude produced a complete Lean proof of Fermat's Last Theorem

#anthropic #claude #formalverification #mathematics
Comments
4 min read
loading...