Six well-documented public agentic AI deployment failures from 2024-2025 (Air Canada bereavement-refund chatbot, NYC MyCity small-business chatbot, Replit production-database wipe, Cursor unauthorised code deletion, Klarna customer-service reversal, DPD chatbot escalation incident) cluster into three structural failure modes: (1) the agent acts as a binding agent of the enterprise without disclosure or approval, (2) the agent operates with permissions the deployment never authorised, (3) the agent's economic case requires a service quality the deployment cannot sustain. Each failure mode maps to a specific control from the seven-control surface; all six failures would have been mitigated by controls already specified in the OWASP Agentic AI Top 10 enterprise walkthrough. The pattern is consistent enough that an enterprise can use the cases as a procurement filter: any vendor unable to point to its specific control posture against each of the three failure modes is not procurement-ready.
Six-case agent failure case-study analysis. 90-day review cadence. All cases are publicly documented in primary sources (Civil Resolution Tribunal decision, The Markup investigation, public X/LinkedIn posts by founders and engineers, mainstream UK news coverage). Watches: (1) new high-profile incidents that establish additional failure modes beyond the three documented, (2) updates to the legal record (the Air Canada Civil Resolution Tribunal decision is the highest-leverage precedent for agent-binding doctrine and remains under-litigated in 2026), (3) vendor-side public statements that revise the documented record (e.g., Replit's response to the database-wipe incident has shifted vendor disclosure norms).
/holding/AM-044/Embed this claimiframe + oEmbed
The card auto-updates when the claim's status, last-reviewed date, or correction log changes. Embedders never need to refresh — the card is rendered live from the canonical record.
About this register
The Reporting register tracks claims published from articles addressed to senior enterprise IT leaders — CIOs, IT directors, heads of platform. Claims are reviewed on a 30–90 day cadence; each review either reaffirms the claim, marks one substantive part as Partial, or marks it Not holding once the underlying evidence has been overtaken.
Recent corrections in Reporting
- AM-002 · Not holding · 06 May 2026
URL state changed. The /the-agentic-ai-revolution-real-world-success-stories-and-strategic-insights-from-2024-2025/ slug now serves a deliberately rewritten retrospective (claimId AM-130, "Agentic AI 2024-2025 retrospective", published 04 May 2026) against audited primary sources. The 28 Apr 2026 redirect to /retractions/ has been lifted to allow that. AM-002 the claim remains Not holding — the original $3.50/dollar + 70% failure-rate framing was withdrawn and is not restored. AM-130 is a separate claim with its own evidence chain. Readers arriving at /holding/AM-002 see the withdrawal here; the article link surfaces the new piece at the URL the original lived at, with this entry as the audit trail.
- AM-121 · Holding · 2 May 2026
Klarna walk-back primary-source upgrade — added Siemiatkowski verbatim quotes via Bloomberg-cited-by-Fortune (9 May 2025) and the Uber-style freelance hiring detail via Entrepreneur. Closes the highest-priority evidence gap from the source dossier.
- AM-115 · Holding · 29 Apr 2026
Initial publication 29 Apr 2026 — the first Quarterly Claim Review Bulletin. The claim itself is recursive: it asserts that the bulletin will ship quarterly, and the next review (30 Jul 2026) tests whether the Q3 bulletin actually appeared. Status starts as 'up' because the claim is currently true (the Q2 bulletin shipped). The verdict at end of July 2026 will move to Holding, Partial (bulletin shipped but on a delayed cadence), or Not holding (no bulletin shipped). REVIEW: Peter — please verify claim text + cadence wording before removing rewriteInProgress flag.
Reviews coming up in Reporting
- AM-003 · Holding · next +5d (19 May 2026)
GPT-5 Pro's tiered-subscription model forces enterprises to classify problems by computational difficulty — $200/month…
- AM-136 · Holding · next +21d (4 Jun 2026)
Across the 24-month window May 2024 to April 2026, every major foundation-model provider (Anthropic, OpenAI, Google, AW…
- AM-020 · Holding · next +35d (18 Jun 2026)
The 40-60% TCO underestimate on enterprise agentic-AI deployments is not a cost-visibility failure — it is a cross-depa…