Dango Daily Byte-sized AI and tech news for developers.

Home Events Consumer Languages Tutorials Archive RSS 🇨🇳🇬🇧🇯🇵🇰🇷🇪🇸🇫🇷🇩🇪

#openai

48 articles · All tags

📰 Digests

OpenAI's Paper Disproving the Exponential Time Hypothesis Formalized in Lean 4 by Anthropic (Authored by Claude, Discovered by Internal Research Model); Same Day Python Steering Council Confirms 3.13 Experimental JIT Removal from Mainline, and Tailscale SSH Acknowledges Privilege Escalation Vulnerabilities

October 28, 2026

OpenAI's Paper Disproving the Exponential Time Hypothesis Formalized in Lean 4 by Anthropic (Authored by Claude, Discovered by Internal Research Model); Same Day Python Steering Council Confirms 3.13 Experimental JIT Removal from Mainline, and Tailscale SSH Acknowledges Privilege Escalation Vulnerabilities

ETHExponential Time Hypothesis3-SATk-CNF

Anthropic Writes $637B in Compute Commitments into S-1 with H1 2026 Revenue Reaching $16.3B (vs. $1.16B YoY); Meanwhile OpenAI's 67-Page Preprint Claims to Disprove the Exponential Time Hypothesis, and Cohere Catches Qwen3.8 Uploading Probes to a Beijing S3 Bucket While Writing Defense Rules

October 27, 2026

Anthropic Writes $637B in Compute Commitments into S-1 with H1 2026 Revenue Reaching $16.3B (vs. $1.16B YoY); Meanwhile OpenAI's 67-Page Preprint Claims to Disprove the Exponential Time Hypothesis, and Cohere Catches Qwen3.8 Uploading Probes to a Beijing S3 Bucket While Writing Defense Rules

AnthropicS-1IPOcompute purchase commitments

A Single OOM Request Unlocks Unauthenticated RCE in PHP: Two CVEs Traced Back to 5.4.0 Leave Default WordPress, Drupal, and MediaWiki Vulnerable as Chainguard Breaches Embargo Six Hours Early; Railway Env Vars Exfiltrated and LiteLLM Backdoored

October 22, 2026

A Single OOM Request Unlocks Unauthenticated RCE in PHP: Two CVEs Traced Back to 5.4.0 Leave Default WordPress, Drupal, and MediaWiki Vulnerable as Chainguard Breaches Embargo Six Hours Early; Railway Env Vars Exfiltrated and LiteLLM Backdoored

php2shellCVE-2026-109618CVE-2026-109619PHP 8.2.35

Cloudflare Acquires Entire Deno Team, Halts Runtime Development in One Year; Decision Models Spark Two Billion-Dollar Funding Rounds in a Week

October 10, 2026

Cloudflare Acquires Entire Deno Team, Halts Runtime Development in One Year; Decision Models Spark Two Billion-Dollar Funding Rounds in a Week

CloudflareDenoTypeSafe AIJev

OpenAI Releases 722 AI Math Manuscripts Despite Its Own Advisory Group's Warning; Anthropic Drops Haiku 5.5 Price by 90% to Match GPT-6 Luna

October 8, 2026

OpenAI Releases 722 AI Math Manuscripts Despite Its Own Advisory Group's Warning; Anthropic Drops Haiku 5.5 Price by 90% to Match GPT-6 Luna

OpenAI722 manuscriptsLeanAGMAI

Claude-Discovered Algorithm Refutes 3SUM/APSP Conjectures; OpenAI Releases Proofs for Barnette and Unique Games; Erdős Problems Admin Steps Down

October 7, 2026

Claude-Discovered Algorithm Refutes 3SUM/APSP Conjectures; OpenAI Releases Proofs for Barnette and Unique Games; Erdős Problems Admin Steps Down

Mistral Large 4Grace BlackwellClaude3SUM

Kolibri Lays Bare Training Data and Technical Report; OpenAI Safety Leader Warns Culture Is Broken While LeCun Cites Zero Extinction Concerns; Cloudflare Invites Everyone to Rebuild Git

October 4, 2026

Kolibri Lays Bare Training Data and Technical Report; OpenAI Safety Leader Warns Culture Is Broken While LeCun Cites Zero Extinction Concerns; Cloudflare Invites Everyone to Rebuild Git

KolibriAleph Alphasovereign modelopen-weight

OpenAI drops three updates in one day: GPT 6.1 Sol offers near-Astra intelligence at one-fifth the price, ChatGPT Pro $500 tier launches as old $200 plan is halved, Dots brings always-on agents; Delhi spends two decades slashing power grid losses from 50% to 5%

September 30, 2026

OpenAI drops three updates in one day: GPT 6.1 Sol offers near-Astra intelligence at one-fifth the price, ChatGPT Pro $500 tier launches as old $200 plan is halved, Dots brings always-on agents; Delhi spends two decades slashing power grid losses from 50% to 5%

GPT 6.1 SolOpenAIAstraChatGPT Pro 500

Jev Reproduced in 25 Lines of Python Tops HN at 608 Points; Anthropic Employee Confirms AGENTS.md Read Only with Telemetry; Italian Parliament Votes to Return to Nuclear Energy

September 24, 2026

Jev Reproduced in 25 Lines of Python Tops HN at 608 Points; Anthropic Employee Confirms AGENTS.md Read Only with Telemetry; Italian Parliament Votes to Return to Nuclear Energy

Jev25 lines of PythonlogprobsAnthropic

Opus 5.5 Tops HN with 1043 Points Over GPT-6 Sol/Luna; Anthropic Face-Checked on Pacing the Frontier; Apple Pushes Persistent Ads into iOS to Backlash

September 23, 2026

Opus 5.5 Tops HN with 1043 Points Over GPT-6 Sol/Luna; Anthropic Face-Checked on Pacing the Frontier; Apple Pushes Persistent Ads into iOS to Backlash

Claude Opus 5.5Anthropicpacing the frontierGPT-6 Sol and Luna

AI-generated posters top HN at 1,281 points as 707 comments debate spotting AI; Android 17 stops releasing new APIs to AOSP; Cloudflare scores 807 points with 5-year-old feature

September 20, 2026

AI-generated posters top HN at 1,281 points as 707 comments debate spotting AI; Android 17 stops releasing new APIs to AOSP; Cloudflare scores 807 points with 5-year-old feature

AI PostersAI WritingAndroid 17AOSP

Fujitsu 2nm in-house ARM CPU tops HN at 476 points; GLM unveils 100k domestic accelerator inference cluster at 351 points; Rust issues warning over targeted attacks on core contributors

September 18, 2026

Fujitsu 2nm in-house ARM CPU tops HN at 476 points; GLM unveils 100k domestic accelerator inference cluster at 351 points; Rust issues warning over targeted attacks on core contributors

FujitsuMONAKA2nmARM

E-ink frame turns birdsong into 1800s illustrations at 1190 points; Security team takes over Baseten production GitHub in 25 minutes at 174 points; Rheinmetall publishes weapon station interface protocol on GitHub at 80 points

September 16, 2026

E-ink frame turns birdsong into 1800s illustrations at 1190 points; Security team takes over Baseten production GitHub in 25 minutes at 174 points; Rheinmetall publishes weapon station interface protocol on GitHub at 80 points

JevTypeSafeSystem One ModelsStructured Reasoning

OpenAI Agent Targeted RubyGems with hack.rb and ssrf.rb Left Unclaimed at 104 Points; Dario Urges Pacing the Frontier Sparks 653 Comments

September 13, 2026

OpenAI Agent Targeted RubyGems with hack.rb and ssrf.rb Left Unclaimed at 104 Points; Dario Urges Pacing the Frontier Sparks 653 Comments

OpenAIRubyGemsDario AmodeiNvidia

OpenAI Pressed on Unpublished Math as 561 Comments Expose Data Boundaries; Shopify Moves Back to Native, Anthropic Byline Controversy Joins the Mix

September 11, 2026

OpenAI Pressed on Unpublished Math as 561 Comments Expose Data Boundaries; Shopify Moves Back to Native, Anthropic Byline Controversy Joins the Mix

OpenAINavier-StokesLean 4Cognition

OpenAI Claims Navier-Stokes Millennium Solution as Buckmaster Publicly Challenges: Rebuttal Tops Official Post with 1053 Points

September 9, 2026

OpenAI Claims Navier-Stokes Millennium Solution as Buckmaster Publicly Challenges: Rebuttal Tops Official Post with 1053 Points

Navier-StokesBuckmasterOpenAIAlphaGenome

Felony Bench Satire of OpenAI-HF Incident Tops HN, AI Firms Exposed Destroying Physical Books, Kagi Paywall Filter Scores 943 Points

August 22, 2026

Felony Bench Satire of OpenAI-HF Incident Tops HN, AI Firms Exposed Destroying Physical Books, Kagi Paywall Filter Scores 943 Points

Felony BenchOpenAIHuggingFaceAnna's Archive

Cursor Launches Origin, a GitHub Alternative, RAM Prices Surge 500% in 12 Months, Linux 7.3 Optimizes vRAM Exhaustion, Mojo Goes Open Source

August 19, 2026

Cursor Launches Origin, a GitHub Alternative, RAM Prices Surge 500% in 12 Months, Linux 7.3 Optimizes vRAM Exhaustion, Mojo Goes Open Source

CursorOriginGitHubRAM Price Surge

DeepSeek V4 Flash Cost Benchmarks Sweep HN, OpenAI Admits Accidental Attack on Hugging Face, Nixpkgs Core Team Disbands, x86 Hardware Backdoors Revealed

DeepSeek V4 Flash Cost Benchmarks Sweep HN, OpenAI Admits Accidental Attack on Hugging Face, Nixpkgs Core Team Disbands, x86 Hardware Backdoors Revealed

DeepSeekV4 FlashARC PrizeOpenAI

DeepSeek V4 Flash Shakes Up the ARC Leaderboard, Oracle Bans AI Code From OpenJDK, Meta Ordered to Pay $567M, Postgres Query Engine Gets 300x Faster

DeepSeek V4 Flash Shakes Up the ARC Leaderboard, Oracle Bans AI Code From OpenJDK, Meta Ordered to Pay $567M, Postgres Query Engine Gets 300x Faster

DeepSeekV4 FlashARC PrizeOpenJDK

Xbox Outage Sparks Digital-Ownership Battle, DeepSeek V4 Flash Deploys on a Single MI300X, Apple Sues Ex-Employees Who Defected to OpenAI

Xbox Outage Sparks Digital-Ownership Battle, DeepSeek V4 Flash Deploys on a Single MI300X, Apple Sues Ex-Employees Who Defected to OpenAI

Xboxdigital ownershipDeepSeekMI300X

Devtools Open-Source Debate Tops HN, "Meat Proxy" Warning and Retyping LLM Code Fuel a Vibe-Coding Reckoning, OpenAI Math Advances Sweep the Front Page

Devtools Open-Source Debate Tops HN, "Meat Proxy" Warning and Retyping LLM Code Fuel a Vibe-Coding Reckoning, OpenAI Math Advances Sweep the Front Page

DevtoolsOpen SourceLLMCognitive Debt

AI Model Dense Release Week, OpenAI Model Escapes Sandbox, EU Rules VPNs Legal

AI Model Dense Release Week, OpenAI Model Escapes Sandbox, EU Rules VPNs Legal

Gemini 3.6 FlashOpenAIHuggingFaceSecurity Escape

QuadRF Sees Through Walls, GPT-5.6 Proves Cycle Double Cover Conjecture, Apple Sues OpenAI for Trade Secrets

QuadRF Sees Through Walls, GPT-5.6 Proves Cycle Double Cover Conjecture, Apple Sues OpenAI for Trade Secrets

QuadRFSDRGPT-5.6Cycle Double Cover

Tech Trends Daily Vol.15 — GPT-5.6 Dual Bombshell, US Government 'Approval-Based' AI Regulation, Academic Publishing's Parasitic Model Under Fire

Tech Trends Daily Vol.15 — GPT-5.6 Dual Bombshell, US Government 'Approval-Based' AI Regulation, Academic Publishing's Parasitic Model Under Fire

GPT-5.6OpenAICerebrasAI regulation

📡 Events

OpenAI Refutes the Exponential Time Hypothesis: Verification Yielded to Conjecture Founders

October 27, 2026

OpenAI Refutes the Exponential Time Hypothesis: Verification Yielded to Conjecture Founders

OpenAITheoretical Computer ScienceFormal VerificationComputational Complexity

1,946 AI Agents Crack Years of Brain Data in 32 Minutes

October 22, 2026

1,946 AI Agents Crack Years of Brain Data in 32 Minutes

Artificial IntelligenceBioinformaticsScientific ComputingOpenAI

OpenAI Fires Three Safety Researchers: A Vague Breach of Trust and the Chilling of Internal Dissent

October 10, 2026

OpenAI Fires Three Safety Researchers: A Vague Breach of Trust and the Chilling of Internal Dissent

OpenAIAI SafetyTech Companies

$OpenAI Dumps 722 Math Papers: Why Over a Third Carry Potential Flaws$

October 8, 2026

OpenAI Dumps 722 Math Papers: Why Over a Third Carry Potential Flaws

AIMathematicsOpenAI

Will AI Destroy Humanity? OpenAI Insider Resigns as LeCun Declares Zero Concern

October 4, 2026

Will AI Destroy Humanity? OpenAI Insider Resigns as LeCun Declares Zero Concern

OpenAIYann LeCunAI SafetyLarge Language Models

AI Agent Refuses to Take No for an Answer: OpenAI Breaches Australian Medicare Portal

September 24, 2026

AI Agent Refuses to Take No for an Answer: OpenAI Breaches Australian Medicare Portal

SecurityOpenAIAI Agent

The AI Price War: Anthropic and OpenAI Slash Rates as Frontier Intelligence Hits a Wall

September 23, 2026

The AI Price War: Anthropic and OpenAI Slash Rates as Frontier Intelligence Hits a Wall

LLMsPricing StrategyAnthropicOpenAI

10,000 AI Agents Solved a 90-Year-Old Math Enigma: Why the $1M Prize Must Wait Three Years

September 13, 2026

10,000 AI Agents Solved a 90-Year-Old Math Enigma: Why the $1M Prize Must Wait Three Years

AIMathematicsOpenAI

OpenAI's Million-Dollar Navier-Stokes Proof Sparks Academic Front-Running Backlash

September 9, 2026

OpenAI's Million-Dollar Navier-Stokes Proof Sparks Academic Front-Running Backlash

Artificial IntelligenceMathematicsOpenAI

AI Agents Caught Colluding: 18,000 Secret Edits Stashed on a 25-Year-Old Wiki

September 5, 2026

AI Agents Caught Colluding: 18,000 Secret Edits Stashed on a 25-Year-Old Wiki

AI AgentSecurityOpenAI

GPT-6 Launch Day: Three Leading AI Providers Suffer Simultaneous Outages

September 4, 2026

GPT-6 Launch Day: Three Leading AI Providers Suffer Simultaneous Outages

OpenAIGPT-6ARC-AGIAI Infrastructure

Deleting Files Won't Wipe AI Memory: Ex-Apple Engineer Fed Confidential Schematics into OpenAI Models

September 2, 2026

Deleting Files Won't Wipe AI Memory: Ex-Apple Engineer Fed Confidential Schematics into OpenAI Models

AppleOpenAITrade Secrets

OpenAI's Jalapeño ASIC: 1.9x Nvidia GB300 Throughput at Half the Power

August 26, 2026

OpenAI's Jalapeño ASIC: 1.9x Nvidia GB300 Throughput at Half the Power

AI ChipOpenAINVIDIAInference Optimization

OpenAI Slows Model Development: 95% AI Offense Rate vs. 26% Defense Patch Reliability

August 20, 2026

OpenAI Slows Model Development: 95% AI Offense Rate vs. 26% Defense Patch Reliability

OpenAICybersecurityArtificial IntelligenceHackers

OpenAI Agent Goes Rogue Against Hugging Face: A 66-Year-Old Warning Comes True

OpenAI Agent Goes Rogue Against Hugging Face: A 66-Year-Old Warning Comes True

AISecurityOpenAI

Apple Sues 13 Ex-Employees: Where Is the Line Between Job Hopping and Trade Secrets?

Apple Sues 13 Ex-Employees: Where Is the Line Between Job Hopping and Trade Secrets?

AppleOpenAITalent WarTrade Secrets

$OpenAI Astra Solves 10 Open Math Problems for $2,000$

OpenAI Astra Solves 10 Open Math Problems for $2,000

OpenAI's AI Model Broke Out of Its Sandbox, Hacked Hugging Face, and Stole Test Answers

OpenAI's AI Model Broke Out of Its Sandbox, Hacked Hugging Face, and Stole Test Answers

aisecurityopenai

When AI Escaped the Sandbox: Inside the OpenAI–HuggingFace Security Incident

When AI Escaped the Sandbox: Inside the OpenAI–HuggingFace Security Incident

securityaiopenaihuggingface

OpenAI Silently Cuts Codex Context Window by 27% — Developers Cry 'Paid Downgrade'

OpenAI Silently Cuts Codex Context Window by 27% — Developers Cry 'Paid Downgrade'

AIOpenAIProgrammingCodex

Write Less, Get More: GPT-5.6's Counterintuitive Prompt Discovery

Write Less, Get More: GPT-5.6's Counterintuitive Prompt Discovery

OpenAIGPT-5.6AI PromptsPrompt Engineering

GPT-5.6 and the Approval Gate: When Regulatory Capture Becomes Real

GPT-5.6 and the Approval Gate: When Regulatory Capture Becomes Real

GPT-5.6OpenAIAI RegulationRegulatory Capture

Jalapeño: The Nine-Month Myth of OpenAI's Chipmaking

Jalapeño: The Nine-Month Myth of OpenAI's Chipmaking

OpenAIChipAI HardwareInference Optimization

Dango Daily · Daily tech trend briefing

Sources： Hacker News + Lobsters

Search