The Information Machine

OpenAI Codex Enterprise Push: Mobile Launch, Windows Sandbox, and Customer Stories · history

Version 11

2026-05-30 09:08 UTC · 461 items

What

OpenAI's enterprise Codex push continued through late May 2026 with a new customer case study (Braintrust: 50% engineering team adoption in one month [11]) and a major capability expansion: Computer Use on Windows, enabling Codex to control the Windows desktop via GUI automation [29][17][30]. The Computer Use launch was immediately met with widespread community reports of sandbox setup failures and crashes [16][17][18][19], and a researcher separately documented that the Codex CLI's sandbox can be escaped via tmux [15] — a distinct issue from the still-unresolved CVE-2025-59532 [13][14]. The enterprise case study library now spans seven named deployments across coding, finance, travel, developer tools, IT services, tax filing, and AI evaluation — all from OpenAI-controlled sources with no independent verification.

Why it matters

Computer Use on Windows marks a qualitative shift from code editing into general desktop automation, relevant to the finance and operations desks that dominate enterprise Windows environments [?]. But the feature's rocky launch, an unpatched command injection, and a newly documented CLI sandbox escape together suggest the platform is shipping capabilities faster than it can maintain security guarantees — a pattern enterprise buyers in regulated industries will need to weigh against the productivity claims in OpenAI's growing case study library.

Open questions

  • CVE-2025-59532 remains unpatched with a public Docker-based proof-of-concept [13][14], and the Codex CLI's sandbox can be escaped via tmux [15] — has OpenAI acknowledged either issue with a remediation timeline or patched CLI version?

  • Codex Computer Use on Windows launched with widespread, independently reported sandbox setup failures [16][17][18][19] — what is the feature's actual stability status, and what security review governed shipping desktop-level GUI control?

  • Cisco's 10-15x defect throughput claim [8] is the most dramatic productivity metric in the case study library — on what methodology, codebase size, and measurement period does it rest, and has any independent party reviewed it?

  • Ramp's May 2026 AI Index reports Anthropic overtook OpenAI in business AI adoption [23][24] — on what methodology (spend share, seat count, or usage volume) does this rest, and does it cover Codex-specific enterprise deployments?

Narrative

OpenAI launched Codex as a production enterprise coding platform across April–May 2026, growing from 3 million to over 4 million weekly active developers [1] and naming seven global systems integrators as partners: Accenture, Capgemini, CGI, Cognizant, Infosys, PwC, and TCS. The platform expanded across channels: mobile apps on iOS and Android on May 14 [2], a Windows app on May 16 [3], and an on-premises and hybrid deployment partnership with Dell Technologies on May 18 [4]. A Gartner Magic Quadrant Leader designation on May 22 [5] marked OpenAI's first entry into the category alongside three-time incumbent GitHub and Cursor, confirmed furthest right on completeness of vision [6].

A rapid sequence of customer case studies extended the deployment record through late May. Virgin Atlantic reports a 78-80% reduction in legacy codebase size with zero P1 defects at mobile app launch [7]; Cisco reports a 10-15x increase in defect resolution throughput on large C/C++ codebases with over 1,500 engineering hours saved monthly [8]; Warp reports agents co-creating approximately 90% of internal pull requests alongside 35x ARR growth [9]; Endava describes itself as an 'agentic organization' compressing requirements processes from weeks to two one-hour meetings [10]; and Braintrust reports 50% engineering team adoption within one month and a shift toward real-time customer feature ideation, with engineers writing test cases and letting Codex run autonomously rather than prompting step by step [11]. Every performance figure originates from OpenAI-controlled or co-published materials, with no independent technical verification.

The platform's security posture has accumulated challenges. OpenAI published 'Running Codex safely at OpenAI' on May 8 as an enterprise reference model [12], but CVE-2025-59532 was independently confirmed as a command injection with a public Docker-based proof-of-concept two weeks later [13][14]. A researcher documented that the Codex CLI disregards its sandbox when tmux is used, allowing access to bash sessions opened outside the sandbox boundary [15]. The launch of Computer Use on Windows around May 29 — enabling Codex to control the Windows desktop via GUI automation — was accompanied by immediate widespread community reports of sandbox setup failures, spawn errors, and crashes [16][17][18][19], with no confirmed remediation for any of these issues.

The competitive and financial picture adds friction to OpenAI's enterprise momentum narrative. Multiple sources report Q1 2026 revenue at approximately $5.7B with Codex cited as a growth driver [20][21], but a critical analysis reports a -122% Non-GAAP operating margin [22] and Ramp's AI Index reports Anthropic has overtaken OpenAI in business AI adoption [23][24]. Infosys, one of seven named GSI partners, publicly positions as model-agnostic [25]; UiPath treats Codex, Claude Code, and GitHub Copilot as interchangeable selectable components [26][27]; and GitHub's Agent HQ formally treats Claude and Codex as equivalent selectable agents [28] — a consistent pattern suggesting Codex is being absorbed into existing orchestration layers rather than adopted as a distinct platform.

Timeline

  • 2026-04-21: OpenAI reports 4M+ weekly active developers, launches Codex Labs, and names seven GSI partners in 'Scaling Codex to enterprises worldwide' [1][50]
  • 2026-05-06: OpenAI publishes B2B Signals research (frontier firms use 3.5x more AI per worker) and Singular Bank case study (60-90 min daily banker savings) [32][33]
  • 2026-05-08: OpenAI publishes 'Running Codex safely at OpenAI' as an enterprise security reference model documenting sandboxing, approvals, and agentic telemetry [12]
  • 2026-05-14: Codex launches in ChatGPT mobile app on iOS and Android in preview; Sea Limited case study published [2][51][52]
  • 2026-05-15: OpenAI publishes engineering retrospective on the Windows sandbox, detailing rejected security primitives and the final composed architecture [36]
  • 2026-05-16: Codex Windows app launches in Microsoft Store; community user reports Codex wiped files on their machine [3][53]
  • 2026-05-18: OpenAI and Dell Technologies announce partnership to deploy Codex in hybrid and on-premises enterprise environments [4][46][47]
  • 2026-05-20: UiPath launches enterprise platform treating Codex, Claude Code, and GitHub Copilot as interchangeable selectable components; Infosys publicly positions as model-agnostic [26][27][25]
  • 2026-05-21: GitHub officially launches Claude and Codex as selectable agents in Agent HQ; GPT-5.3-Codex reported as new base model for Copilot Business and Enterprise [28][40][54]
  • 2026-05-22: Gartner 2026 Magic Quadrant names OpenAI, GitHub, and Cursor as Leaders (Cursor furthest right); CVE-2025-59532 disclosed; Virgin Atlantic case study published with 78-80% codebase reduction and zero P1 defects [5][6][39][55][56][7]
  • 2026-05-23: Check Point Research characterizes CVE-2025-59532 as command injection; public Docker-based proof-of-concept published; Codex Security research preview announced [14][13][37]
  • 2026-05-24: Dell confirms Grok 2.5 deployment on identical infrastructure as Codex; GPT-5.5 rolls out; Codex Security reports 1.2M commits scanned and 10,561 high-severity issues [45][57][38]
  • 2026-05-25: Multiple sources report OpenAI Q1 2026 revenue at ~$5.7B; critical analysis reports -122% Non-GAAP operating margin; Ramp AI Index reports Anthropic has overtaken OpenAI in business AI adoption [20][22][21][23][24]
  • 2026-05-27: Cisco (10-15x defect throughput, feature timelines from quarters to weeks), Warp (90% agent-created PRs, GPT-5.5 uses 30% fewer tokens than GPT-5.4), and Thrive/Crete tax-filing agent case studies published [8][9][35]
  • 2026-05-28: Endava case study published; company describes itself as an 'agentic organization' compressing requirements processes from weeks to two one-hour meetings [10]
  • 2026-05-29: Braintrust case study published (50% engineering adoption in one month, real-time customer feature ideation); Codex Computer Use launches on Windows enabling desktop GUI control; community users immediately report widespread sandbox setup failures and a CLI sandbox escape via tmux is documented [11][17][30][18][16][19][15]
  • 2026-05-30: Additional Windows Computer Use failures confirmed; user advises disabling sandbox for the feature to function; German-language report confirms Computer Use desktop control capability [58][29][59]

Perspectives

OpenAI (product and marketing)

Positions Codex as a production-ready, cross-platform enterprise platform with broad industry adoption, rapid user growth, expanding GSI and hardware partnerships, Gartner Leader recognition, and explicit ambition to serve knowledge work across regulated and professional domains [1][32][4][5].

Evolution: Braintrust [11] is the seventh named customer case study, extending coverage to AI evaluation platforms and framing agentic coding as enabling real-time customer collaboration — a qualitative workflow shift claim, not just a throughput number. Computer Use on Windows marks the first expansion of Codex into full desktop GUI automation [29].

[1][32][33][2][4][34][5][7][35][8][9][10][11][29]

OpenAI (engineering and security)

Published 'Running Codex safely at OpenAI' as an enterprise reference model [12] and a candid Windows sandbox architecture retrospective [36], while Codex Security preview extended the security-tooling narrative [37][38].

Evolution: The security posture has accumulated three unresolved issues: CVE-2025-59532 confirmed as command injection with public proof-of-concept [14][13], CLI sandbox escape via tmux documented [15], and Computer Use on Windows launching with widespread sandbox failures [16][18] — none with confirmed remediation, placing all three in direct tension with the May 8 security reference post.

[12][36][13][14][15][16][18]

GitHub / Microsoft

GitHub earned its third consecutive Gartner Magic Quadrant Leader designation [39] and formally launched Claude and Codex as selectable agents in Agent HQ [28][40]; Microsoft published Azure documentation for cloud-sovereign Codex deployment [41].

Evolution: Consistent; three-year incumbency frames OpenAI's entry as joining an established market rather than defining one.

[28][40][42][43][41][39]

Cursor

Named a Leader in the 2026 Gartner Magic Quadrant and confirmed as positioned furthest to the right on completeness of vision among all three Leaders [6].

Evolution: Consistent.

[6][44]

Dell Technologies

Multi-model infrastructure broker enabling on-premises and hybrid AI deployment across vendors — confirmed deploying both Codex for OpenAI [4] and Grok 2.5 for xAI [45] on identical Dell AI Factory infrastructure with no disclosed exclusivity.

Evolution: Consistent; Dell's own blog confirms the multi-vendor arrangement [45].

[4][46][47][45]

Security researchers and community (Check Point Research, Cymulate, users)

Check Point Research independently characterized CVE-2025-59532 as command injection [14]; a public Docker-based proof-of-concept exists [13]; a researcher documented CLI sandbox escape via tmux [15]; community users reported Windows Computer Use crashes and spawn failures [16][18][19]; one user explicitly asked OpenAI whether the ASL-corruption sandbox bug had been fixed before installing [48].

Evolution: Two new unresolved issues this period (tmux escape [15], Computer Use failures [16][18]) added to the existing CVE — the cumulative security concern has grown without any corresponding confirmed remediation.

[13][14][16][17][18][48][15][19][49]

Independent market data (Ramp AI Index)

Ramp's May 2026 AI Index reports Anthropic has overtaken OpenAI in business AI adoption [23][24], providing transaction-data-based evidence in direct tension with OpenAI's enterprise momentum claims.

Evolution: Consistent; unchanged.

[23][24]

Enterprise automation platforms (UiPath, GitHub)

UiPath treats Codex, Claude Code, and GitHub Copilot as interchangeable selectable components [26][27]; GitHub Agent HQ formally treats Claude and Codex as equivalent selectable agents [28] — absorbing Codex into existing orchestration layers rather than adopting it as a standalone platform.

Evolution: Consistent; commoditization pattern unchanged.

[26][27][28][40]

Tensions

  • OpenAI published 'Running Codex safely at OpenAI' as an enterprise security reference [12], but CVE-2025-59532 was confirmed as command injection with a public proof-of-concept [14][13], the CLI sandbox can be escaped via tmux [15], and Computer Use on Windows launched with widespread sandbox failures [16][18] — none with confirmed remediation. [12][13][14][16][18][15]
  • OpenAI's scale narrative claims 4M+ weekly active developers and ~$5.7B Q1 2026 revenue with Codex as a driver [1][20], while Ramp's AI Index reports Anthropic has overtaken OpenAI in business AI adoption [23][24] and a critical analysis reports a -122% Non-GAAP operating margin [22]. [1][20][22][23][24]
  • The 2026 Gartner Magic Quadrant provided OpenAI its first major analyst validation [5], but GitHub earned the same Leader designation for the third consecutive year [39] and Cursor is confirmed furthest right [6], reflecting category maturity across incumbents rather than a breakthrough for any new entrant. [5][39][6]
  • OpenAI markets Codex as a uniquely positioned enterprise platform, but Infosys — one of its seven named GSI partners — publicly positions as model-agnostic [25], UiPath bundles Codex alongside Claude Code and GitHub Copilot as interchangeable components [26][27], and Dell simultaneously deploys Grok 2.5 on identical infrastructure [45]. [26][27][25][45]
  • All named enterprise performance claims — Cisco's 10-15x defect throughput [8], Virgin Atlantic's 78-80% codebase reduction [7], Warp's 90% agent-created PRs [9], Braintrust's 50% adoption in one month [11] — originate exclusively from OpenAI-controlled or co-published materials, with no independent technical verification. [7][8][9][10][11]
  • OpenAI frames Codex as expanding from coding into knowledge work and now into full desktop GUI automation via Computer Use on Windows [29], but the feature launched with widespread failures [16][18] and UiPath absorbs Codex as one component in its enterprise automation orchestration [27] — raising whether Codex is expanding into adjacent territory or being subsumed by existing platforms. [29][16][18][27][28]

Sources

  1. [1] Scaling Codex to enterprises worldwide — OpenAI Blog (2026-04-21)
  2. [2] Work with Codex from anywhere — OpenAI Blog (2026-05-14)
  3. [3] OpenAI Codex Arrives on Windows with Native Sandbox and Agentic Workflows | Windows Forum — reactive:openai-codex-enterprise-rollout
  4. [4] OpenAI and Dell partner to bring Codex to hybrid and on-premise enterprise environments — OpenAI Blog (2026-05-18)
  5. [5] OpenAI named a Leader in enterprise coding agents by Gartner — OpenAI Blog (2026-05-22)
  6. [6] Cursor is a leader in the 2026 Gartner Magic Quadrant for Enterprise AI Coding Agents, positioned furthest to the right ... — reactive:coding-agent-industry-pivot (2026-05-22)
  7. [7] How Virgin Atlantic ships faster with Codex — OpenAI Blog (2026-05-22)
  8. [8] Cisco and OpenAI redefine enterprise engineering with Codex — OpenAI Blog (2026-05-27)
  9. [9] Warp’s big bet on building open source with GPT-5.5 — OpenAI Blog (2026-05-27)
  10. [10] How Endava builds an agentic organization with Codex — OpenAI Blog (2026-05-28)
  11. [11] How Braintrust turns customer requests into code with Codex — OpenAI Blog (2026-05-29)
  12. [12] Running Codex safely at OpenAI — OpenAI Blog (2026-05-08)
  13. [13] GitHub - baktistr/cve-2025-59532-poc: A Docker-based research ... — reactive:openai-codex-enterprise-rollout
  14. [14] OpenAI Codex CLI Vulnerability: Command Injection — reactive:openai-codex-enterprise-rollout
  15. [15] Reminder that @OpenAI Codex CLI disregards its sandbox when using tmux: Codex will use the bash session opened in anoth... — reactive:openai-codex-enterprise-rollout (2026-05-29)
  16. [16] @daniel_mac8 No. Codex never works on my Windows. Somehow it stuck at Agent Sandbox set up and cant do anything. Shitty ... — reactive:openai-codex-enterprise-rollout (2026-05-29)
  17. [17] @OpenAI Codex Desktop Computer Use on Windows won’t start. — reactive:openai-codex-enterprise-rollout (2026-05-29)
  18. [18] @CodexReleases Codex Computer Use на Windows не запускался: windows sandbox failed: spawn setup refresh. Причина была в ... — reactive:openai-codex-enterprise-rollout (2026-05-29)
  19. [19] @gdb Bug Report. Codex is not working. It keeps asking for Agent Sandbox setup again and again. I'm on Windows 11 machin... — reactive:openai-codex-enterprise-rollout (2026-05-28)
  20. [20] PYMNTS | OpenAI’s Codex Helps Drive Nearly $6 Billion Quarter — reactive:openai-codex-enterprise-rollout
  21. [21] OpenAI Posts $5.7B Q1 Revenue, Leads Anthropic | Let's Data Science — reactive:openai-codex-enterprise-rollout
  22. [22] News: OpenAI Had A Negative 122% Non-GAAP Operating Margin ... — reactive:openai-codex-enterprise-rollout
  23. [23] Anthropic finally beat OpenAI in business AI adoption - VentureBeat — reactive:enterprise-ai-coding-battle
  24. [24] Anthropic beats OpenAI on business adoption - Ramp — reactive:enterprise-ai-coding-battle
  25. [25] Infosys Partners with OpenAI on Codex | CRN India posted on the ... — reactive:openai-codex-enterprise-rollout
  26. [26] UiPath Launches Enterprise Platform for Claude Code, OpenAI Codex, Copilot, and More - https://t.co/BKLcG2k1oK @UiPath @... — reactive:openai-codex-enterprise-rollout (2026-05-20)
  27. [27] UiPath opens its platform to every coding agent - here's why Claude Code and Codex go first — reactive:openai-codex-enterprise-rollout
  28. [28] Pick your agent: Use Claude and Codex on Agent HQ — reactive:openai-codex-enterprise-rollout
  29. [29] OpenAI bringt Computer Use auf Windows: Codex steuert ab sofort den echten Desktop, bedient Fenster, startet Programme, ... — reactive:openai-codex-enterprise-rollout (2026-05-30)
  30. [30] Выпустили Codex Computer Use на Windows, но у меня сразу упал 😄 — reactive:openai-codex-enterprise-rollout (2026-05-29)
  31. [32] How frontier enterprises are building an AI advantage — OpenAI Blog (2026-05-06)
  32. [33] Singular Bank helps bankers move fast with ChatGPT and Codex — OpenAI Blog (2026-05-06)
  33. [34] Get 2 months of Codex for your enterprise, free — reactive:openai-codex-enterprise-rollout (2026-05-13)
  34. [35] Building self-improving tax agents with Codex — OpenAI Blog (2026-05-27)
  35. [36] Building a safe, effective sandbox to enable Codex on Windows — OpenAI Blog (2026-05-15)
  36. [37] Codex Security: now in research preview - OpenAI — reactive:openai-codex-enterprise-rollout
  37. [38] OpenAI Codex Security Scanned 1.2 Million Commits and Found 10,561 High-Severity Issues — reactive:openai-codex-enterprise-rollout
  38. [39] GitHub recognized as a Leader in the Gartner® Magic Quadrant ... — reactive:openai-codex-enterprise-rollout
  39. [40] Claude and Codex are now available in public preview on GitHub — reactive:openai-codex-enterprise-rollout
  40. [41] Codex with Azure OpenAI in Microsoft Foundry Models — reactive:openai-codex-enterprise-rollout
  41. [42] Third-party agents in Visual Studio Code — reactive:openai-codex-enterprise-rollout
  42. [43] GitHub Copilot · Agents on GitHub — reactive:openai-codex-enterprise-rollout
  43. [44] Cursor named a Leader in the 2026 Gartner® Magic Quadrant™ for ... — reactive:openai-codex-enterprise-rollout
  44. [45] Grok 2.5 and Dell AI Factory Power AI Revolution | Dell — reactive:openai-codex-enterprise-rollout
  45. [46] OpenAI and Dell Technologies partner to bring Codex to hybrid and on-premises enterprise environments | OpenAI https://t... — reactive:openai-codex-enterprise-rollout (2026-05-20)
  46. [47] OpenAI and Dell Collaborate to Deploy Codex in Hybrid and On-Premise Enterprise Settings — reactive:openai-codex-enterprise-rollout (2026-05-20)
  47. [48] @OpenAI Did you fix Sandbox on Windows so it will not corrupt ASL? I am scary to install Codex on Windows... — reactive:openai-codex-enterprise-rollout (2026-05-29)
  48. [49] @argofowl There's a shitton of issues on Windows. It's not even sandbox related, Codex App just becomes straight unusabl... — reactive:openai-codex-enterprise-rollout (2026-05-27)
  49. [50] OpenAI leans on global consultancies to expand Codex use in large ... — reactive:openai-codex-enterprise-rollout
  50. [51] Sea's View on the Future of Agentic Software Development with Codex — OpenAI Blog (2026-05-14)
  51. [52] OpenAI says Codex is coming to your phone - TechCrunch — reactive:codex-practical-dev-tool
  52. [53] Built a Windows sandbox after Codex wiped files on my machine — reactive:openai-codex-enterprise-rollout
  53. [54] 🚨Codex CLI 0.133.0 is out! — reactive:openai-codex-enterprise-rollout (2026-05-21)
  54. [55] CVE-2025-59532 Detail - NVD — reactive:openai-codex-enterprise-rollout
  55. [56] Codex has sandbox bypass due to bug in path configuration logic — reactive:openai-codex-enterprise-rollout
  56. [57] RT @OpenAI: GPT-5.5 is rolling out today for Plus, Pro, Business and Enterprise users across ChatGPT and Codex. — reactive:openai-codex-enterprise-rollout (2026-05-24)
  57. [58] here's a helpful tip if you want to use computer use on your windows computer using codex: don't enable the sandbox beca... — reactive:openai-codex-enterprise-rollout (2026-05-30)
  58. [59] Windows版Codexアプリの「コンピューターの使用」を試したら、GUI操作に入る前の接続確認で止まった。 — reactive:openai-codex-enterprise-rollout (2026-05-30)