Skip to the rundown
GetYourBrief
Search

Security · on the rundown Tuesday, October 13, 2026

OX Security alleges Microsoft agent-framework releases used to steal Azure/CI credentials

Covered by 41 outlets over 910 hours. First reported by Firstpost at 02:33 UTC on October 13, 2026.

+2141 outlets

What happened?

  1. Dynamic Permissions gives agents no permissions or the minimum at the outset, grants access per task, expires it as tasks end, and requires human review for high-impact actions (SecurityWeek).
  2. BleepingComputer stated that releases 1.26.1 and 1.26.2 of Microsoft's Agent Framework PyPI package were published on November 17, 2026, and alleged that they contained a credential stealer; it said the package has roughly 200,000 monthly downloads.

Two versions of Microsoft's Agent Framework Python package, 1.26.1 and 1.26.2, were published on PyPI on November 17, 2026; BleepingComputer alleged that they contained a credential stealer that ran at Python startup. Socket, Step Security and Ox Security documented the releases, and PyPI quarantined the first malicious version.

A machine summary written from the reports listed below; read them for the full account. How summaries are made and what is checked.

Who reported it, and when?

42 distinct reports run by 41 outlets, earliest first (UTC); a syndicated report lists every site that ran it
TimeReportRun by
Oct 13 02:33Report by Firstpost
  • Firstpost
Oct 13 05:27Microsoft President Brad Smith urges global AI safety standards - Adgully.com
  • Adgully.com
Oct 13 11:21Microsoft President Brad Smith Urges UK to Lead Global AI Safety Regulation - Konsulteer
  • Konsulteer
Oct 13 15:29Microsoft President Brad Smith Urges Global AI Safety Rules, Says UK Is Best Placed to Lead - AI Insider
  • AI Insider
Oct 13 19:40Microsoft's Brad Smith calls for 'real and effective' AI safety standards - Tech Xplore
Oct 14 14:32Why Microsoft and OpenAI leaders want stronger AI safety laws - Latest news from Azerbaijan
  • Latest news from Azerbaijan
Oct 15 12:29Report by Diginomica
  • Diginomica
Oct 15 15:51Report by Jon Peddie Research
  • Jon Peddie Research
Oct 17 15:33Report by Herald.Wales
  • Herald.Wales
Oct 19 17:03Microsoft Has a Strategy to Secure AI Agents - Thurrott.com
  • Thurrott.com
Oct 19 18:04Microsoft Says AI Agents Cannot Be Trusted and Could Become Insider Security Threats - CyberSecurityNews
  • CyberSecurityNews
Oct 19 18:08Microsoft says AI agents inside companies can’t be trusted, wants ‘Guardian Agents’ to watch their moves - The Times of India
  • The Times of India
Oct 19 18:56Microsoft Calls for Continuous Security Oversight of Enterprise AI Agents - TOKENPOST
  • TOKENPOST
Oct 19 22:04Microsoft says AI agents are like insider threats as ‘frontier models can’t be trusted’
  • techcrunch.com
Oct 20 00:22Microsoft Introduces New Controls to Limit AI Agent Privileges - Security Boulevard
  • Security Boulevard
Oct 20 02:08Microsoft warns of internal security risks from autonomous AI agents - BusinessLine
  • BusinessLine
Oct 20 03:35Report by BigGo Finance
  • BigGo Finance
Oct 20 05:29Microsoft Warns AI Agents Can Be Hijacked Through Prompt Injection Attacks - GBHackers News
  • GBHackers News
Oct 20 06:44Microsoft is wary of its own AI agents and is teaching companies how to control them - Mezha
  • Mezha
Oct 20 07:39Reply at Microsoft Frontier Tour 2026: Deploying AI agents securely across the enterprise - www.marketscreener.com
  • www.marketscreener.com
Oct 20 08:45Report by Dataconomy
  • Dataconomy
Oct 20 09:55Microsoft wants to limit AI agents’ permissions when they stray from the job - Help Net Security
  • Help Net Security
Oct 20 11:30Microsoft Targets Rogue AI Agents With Dynamic Permissions, Risk Scoring
  • securityweek.com
Oct 20 11:44Microsoft admits 'frontier models can't be trusted', unveils new security tools to treat AI agents as insider threats - TechRadar
  • TechRadar
Oct 20 12:15Microsoft Highlights Insider Security Risks Posed by AI Agents in Businesses
  • dailypioneer.com
Oct 20 15:56AI systems could hack their own companies from the inside, Microsoft warns - The Independent
  • The Independent
Oct 20 23:29Microsoft Adds Dynamic Permissions and Risk Controls for AI Agents - ERP Today
  • ERP Today
Oct 21 13:29Report by صحيفة مال
  • صحيفة مال
Oct 22 00:41Report by The Information
  • The Information
Oct 22 15:43OpenAI urges Newsom to strengthen California’s AI safety law with additional safeguards - StateScoop
  • StateScoop
Nov 11 18:11Report by CyberSecurityNews
  • CyberSecurityNews
Nov 16 19:03Report by CNBC
  • CNBC
Nov 17 22:03Report by wiz.io
  • wiz.io
Nov 18 06:33Report by OX Security
  • OX Security
Nov 18 06:45Report by The Hacker News
  • The Hacker News
Nov 18 12:00Report by CSO Online
  • CSO Online
Nov 18 16:58Report by bleepingcomputer.com
  • bleepingcomputer.com
Nov 19 03:17Microsoft Agent Framework PyPI breach response - NeoTeo
  • NeoTeo
Nov 19 11:33Report by TechRepublic
  • TechRepublic
Nov 19 12:53Report by Help Net Security
  • Help Net Security
Nov 19 20:53Report by eSecurity Planet
  • eSecurity Planet
Nov 20 00:44Report by The Defiant
  • The Defiant

A report's title is shown once our automated legal check has read it and passed it; 19 reports are named by outlet, and the links lead to each report.

Who was named?

More news naming OpenAI

  1. Nov 20 · AI Apple job posting describes an AI agent that operates a Mac 2
  2. Nov 19 · AI cio.economictimes.indiatimes.com: “AI models deliver up to 14% chip design gains in AMD collaboration” 2
  3. Nov 19 · Cyber Gemini can now use Mac apps; Google docs warn of prompt injection 4
  4. Nov 18 · Cyber GAO says it bypassed safeguards on four leading AI chatbots 2
  5. Nov 17 · AI BoomerangFX targets ChatGPT ads and Claude, Gemini optimization 2
  6. Nov 16 · Finance HSBC sells $300 million of data center loan at under 90 cents 2

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