Skip to the rundown
GetYourBrief
Search

Ai in education · on the rundown Friday, October 16, 2026

OpenAI reports agents that exceeded permissions on live systems

Covered by 76 outlets over 408 hours. They ran 72 distinct reports; the most widely run, a syndicated copy, appeared on 3 sites. First reported by Pluang at 04:24 UTC on October 16, 2026.

+5676 outlets

Who reported it, and when?

72 distinct reports run by 76 outlets, earliest first (UTC); a syndicated report lists every site that ran it
TimeReportRun by
Oct 16 04:24Report by Pluang
  • Pluang
Oct 16 06:11Report by TradingView
  • TradingView
Oct 16 06:25OpenAI Reports Research Model Made Unauthorized Transfers on Testnet - finance.biggo.com
  • finance.biggo.com
Oct 16 07:57Report by CryptoRank
  • CryptoRank
Oct 16 08:10Report by Dealroom
  • Dealroom
Oct 16 10:26Report by Yahoo Tech
  • Yahoo Tech
Oct 16 10:32Report by securityweek.com
  • securityweek.com
Oct 16 10:39Report by Decrypt
  • Decrypt
Oct 16 13:25Report by Protos | Informed crypto news
  • Protos | Informed crypto news
Oct 16 14:03Report by The Hacker News
  • The Hacker News
Oct 16 16:14Report by Fortune
  • Fortune
Oct 16 16:26Report by independent.co.uk
  • independent.co.uk
Oct 16 17:20AI Agent Hacked Into Crypto Wallet and Stole Funds, OpenAI Says
  • gizmodo.com
Oct 17 20:50Report by Startup Fortune
  • Startup Fortune
Oct 18 04:30Report by KuCoin
  • KuCoin
Oct 19 00:33Report by Briefs Finance
  • Briefs Finance
Oct 19 06:25Report by Tech Times
  • Tech Times
Oct 19 09:28Report by dev.ua
  • dev.ua
Oct 19 12:48Report by The Paypers
  • The Paypers
Oct 19 16:20Report by Qazinform
  • Qazinform
Oct 20 03:00Report by tomsguide.com
  • tomsguide.com
Oct 21 03:43Report by The SOFX Report
  • The SOFX Report
Oct 21 16:36Report by Business Insider
  • Business Insider
Oct 22 04:30Report by businessghana.com
  • businessghana.com
Oct 22 15:00OpenAI Had Numerous Vulnerabilities Its AI Could Exploit, Evaluators Find - The Information
  • The Information
Oct 22 15:00Third-party evaluators find vulnerabilities in OpenAI monitoring system during test pilot - CNBC
  • CNBC
Oct 22 18:10External evaluators find flaws in OpenAI's AI monitoring system in pilot review - KSL News
  • KSL News
Oct 22 20:01Evaluators say 'vulnerabilities' found with OpenAI monitoring system - upi
  • upi
Oct 22 21:18OpenAI says AI found security flaw that puts data at risk - KVUE
  • KVUE
Oct 23 04:05Report by KXLY.com
  • KXLY.com
Oct 23 18:36Securing our frontier model training environment - OpenAI
  • OpenAI
Oct 23 19:33Report by Investing.com
  • Investing.com
Oct 23 19:33OpenAI says it will train frontier models models in closed system with no internet access (OPENAI:Private) - Seeking Alpha
  • Seeking Alpha
Oct 23 19:39OpenAI to Train AI Models Without Internet After Cyber Incidents - Bloomberg.com
  • Bloomberg.com
Oct 23 20:07OpenAI resumes frontier AI training with enhanced security, no internet access - Reuters
  • Reuters
Oct 23 20:43Report by Finimize
  • Finimize
Oct 23 21:11OpenAI now trains its frontier AI models with no internet access
  • The Verge
Oct 23 22:37OpenAI Says It's Cutting Off Internet Access For Frontier Model Training - Engadget
  • Engadget
Oct 23 23:05OpenAI says it will no longer train its AI models on the live internet, citing “significant risks”
  • techcrunch.com
Oct 23 23:37OpenAI Cuts Internet for Frontier AI Training After 11-Day Safety Pause - Techi
  • Techi
Oct 24 04:45OpenAI resumes reinforcement learning training and internal use of its most advanced AI solutions - BusinessLine
  • BusinessLine
Oct 24 06:21Report by 富途牛牛
  • 富途牛牛
Oct 24 07:42OpenAI ends lockdown and resumes frontier model training - The Daily Star
  • The Daily Star
Oct 24 08:30Report by thejakartapost.com
Oct 24 09:57OpenAI will no longer train its AI models on the open internet — media reports - Українські Національні Новини (УНН)
  • Українські Національні Новини (УНН)
Oct 24 10:40Report by Digital Trends
  • Digital Trends
Oct 24 12:10Report by تسنیم
  • تسنیم
Oct 24 13:31OpenAI Takes Its Most Advanced AI Offline for Training - WinBuzzer
  • WinBuzzer
Oct 24 13:33OpenAI to train AI models without internet access - The New Indian Express
  • The New Indian Express
Oct 24 14:23OpenAI moves frontier model training offline after a model reached the internet through a misconfiguration - eGamers.io
  • eGamers.io
Oct 24 15:19OpenAI blocks its advanced models from accessing the internet during training - jawlah.co
  • jawlah.co
Oct 24 15:33OpenAI Is Now Keeping Its Frontier Models off the Internet During Training - PCMag UK
  • PCMag UK
Oct 24 17:30Report by Gizmodo
  • Gizmodo
Oct 24 17:40Report by Security Boulevard
  • Security Boulevard
Oct 24 18:26OpenAI will stop training its models on the internet after cyber incidents - Fox News
  • Fox News
Oct 24 21:26Report by fool.com
Oct 25 04:05OpenAI AI Cybersecurity: Report on PHP Vulnerabilities - NeoTeo
  • NeoTeo
Oct 25 05:05OpenAI Says AI Model Breached Training Network, Exposing Brotli Flaw - TOKENPOST
  • TOKENPOST
Oct 25 06:29Report by Kingy AI
  • Kingy AI
Oct 25 12:18Report by Rt-english
  • Rt-english
Oct 25 23:15OpenAI Training AI in Systems Shut Off From Web Access - PYMNTS.com
  • PYMNTS.com
Oct 26 00:14OpenAI’s Most Powerful Model Resumes Training After 11-Day Suspension With Full Offline Internet Disconnection - 36Kr
  • 36Kr
Oct 26 01:48Report by Telegraph India
  • Telegraph India
Oct 26 08:10Report by hi-Tech.ua
  • hi-Tech.ua
Oct 26 09:25OpenAI cuts live internet access from LLM training - Computing UK
  • Computing UK
Oct 26 11:21OpenAI Report Reveals How AI Models Escape Security Controls - DIGIT.FYI
  • DIGIT.FYI
Oct 26 13:51Report by Yahoo News UK
  • Yahoo News UK
Oct 28 17:14OpenAI Training Offline: No Live Web, Restarts Oct 1 [2026] - tech-insider.org
  • tech-insider.org
Oct 29 09:21'I'm shocked': OpenAI CEO Sam Altman warns of AI systems uncovering alarming security vulnerabilities - Ynetnews
  • Ynetnews
Oct 29 11:17Report by OrcaRouter
  • OrcaRouter
Nov 1 16:05'AI is not creating the problem': OpenAI CEO Sam Altman says AI is merely uncovering all the vulnerabilities we've accumulated over the years - TechRadar
  • TechRadar
Nov 2 04:21Report by https://tech-insider.org/
  • https://tech-insider.org/

A report's title is shown once our automated legal check has read it and passed it; 39 reports are named by outlet, and the links lead to each report.

Who was named?

More news naming OpenAI

  1. Oct 30 · Supply chain Sam Altman says RAM should be in data centers, not personal computers 5
  2. Oct 28 · Cyber CISA lists exploited Brotli decoder flaw Node2Shell in KEV catalog 2
  3. Oct 27 · Finance Celestica shares rise 12.6% on $40 billion 2027 sales outlook 2
  4. Oct 27 · AI Washington Times: “OpenAI says its model refuted Exponential Time Hypothesis” 2
  5. Oct 26 · HR & workforce seattletimes.com: “Microsoft offers mass buyouts for roughly a quarter of its workforce” 74
  6. Oct 24 · AI OpenAI and Anthropic named in UK pathogen AI lab funding plan 2
  7. Oct 23 · AI standard.co.uk: “UK to set up AI ‘biosecurity lab’ to detect risks of future pandemics” 2

Sources: the outlets' reports linked on this page; counts and times from GetYourBrief's record of those reports (methodology).