Who Gives A Crap suspended an AI email agent after it told a customer their price would double
In July 2026, the direct-to-consumer brand Who Gives A Crap suspended the AI tool it uses to draft some customer emails after the agent told a subscriber their toilet paper delivery would change from 48 rolls at $66.00 to 24 rolls at $69.50, effectively doubling the per-roll price. A second AI-generated email reaffirmed the incorrect figures. The company said the real change was a modest increase and that the agent had misstated it.
Records by entity: Who Gives A Crap
The AI told a customer their subscription would shrink and cost more, then a second AI email confirmed the mistake.
Key facts
- What
- In July 2026, the direct-to-consumer brand Who Gives A Crap suspended the AI tool it uses to draft some customer emails after the agent told a subscriber their toilet paper delivery would change from 48 rolls at $66.00 to 24 rolls at $69.50, effectively doubling the per-roll price.
- Incident date
- Jul 8, 2026
- Who
- Who Gives A Crap
- Failure mode
- Hallucination
- AI surface
- Agentic Workflow
- Severity
- Low
What happened
Who Gives A Crap told customers it was raising prices modestly across its range. One subscriber instead received an AI-drafted email stating their subscription would change from 48 rolls at $66.00 to 24 rolls at $69.50, which halved the quantity and roughly doubled the price per roll. When the customer queried it, the company's support flow sent a second email, also AI-generated, that reaffirmed the wrong numbers and apologized for "unclear messaging." A spokesperson later said the original email contained a typo, the correct price covered 48 rolls, and the company had shut down the email agent and sent a correction once it identified the root cause.
What broke inside the model
- 01 · TriggerA user asks for a fact, a citation, or a figure.
- 02 · Model stepThe model writes a fluent, confident answer.
- 03 · Control gapNothing ties the claim back to a real source.
- 04 · FailureA fabricated fact ships as if it were verified.
- 05 · ConsequenceThe false claim reaches a customer, a court, or the public.
Confidence holds, and even spikes, as the claim detaches from any source.
The agent generated customer-facing pricing language without a validation step tying its output to the actual product and price record. It produced confident, specific, wrong figures, and because the escalation path also routed through the same generative agent, the follow-up reinforced the error rather than catching it. Nothing in the loop checked the numbers against the source of truth before the message left.
What it cost
Sources
- PressWho Gives A Crap suspends AI agent after email error said prices would doublesmartcompany.com.au
Cite this entry
https://failureindex.ai/failures/who-gives-a-crap-ai-email-agent-wrong-priceAI Failure Index. "Who Gives A Crap suspended an AI email agent after it told a customer their price would double" (FI-0702). Realm Labs. https://failureindex.ai/failures/who-gives-a-crap-ai-email-agent-wrong-price (indexed Jul 10, 2026).Data fields CC-BY 4.0, prose citation permitted. Incident ID FI-0702. Full dataset at /data.
Note from Realm Labs, the Index steward
How Realm would have caught this
- Prism
- OmniGuard
- AI Detection & Response (AIDR)
Realm compares the agent's drafted output against the governing data (here, the customer's actual plan and approved pricing) at runtime and blocks or flags a message that asserts figures the record does not support. The same policy gate on the escalation step would have stopped the second email from repeating an unverified claim.