Evidence: Claude Pro Max hallucinated $270 Notion feature - 23 days of silence from Anthropic

README.md

Claude Pro Max $270 Hallucination Evidence

Summary

Claude Pro Max ($200/month) confidently instructed me to add it as a Notion workspace member - a feature that doesn't exist. This cost me $270 in unnecessary Notion subscription.

Timeline of Support Requests

• June 17, 2025: First email → No response
• July 5, 2025: Second email (18 days later) → No response
• July 6, 2025: Escalation → No response
• July 9, 2025: Final ultimatum → Bot reply only
• Total: 23 days of complete silence

Financial Impact

• Claude Pro Max subscription: $807 (3 months)
• Lost to false advice: $270 (Notion annual)
• Total loss: $1,077

Key Evidence

1. Claude's confident but false technical instructions
2. Notion support confirming this is impossible
3. Payment receipts showing financial loss
4. Complete timeline of ignored support requests

About Me

Verified tech innovator who developed GiveCon, pioneering the $3B K-POP fandom app market. 32.6K YouTube subscribers. Not a random complainer but a serious user trying to use AI for creative collaboration.

claude-false-advice.md

Claude's False Technical Instructions

The Question

"How can I integrate Claude with Notion for collaborative book writing?"

Claude's Response (Summary)

1. "Add Claude as a Notion workspace member"
2. "This will enable unlimited document processing"
3. "Enable 'Share to web' feature"
4. "Use web_fetch to access Notion pages"
5. "Claude can then edit documents in real-time"

The Reality

• Notion Support Response: "AI cannot be added as workspace members. This is technically impossible."
• Result: Purchased unnecessary 2-person Notion subscription
• Cost: $270 annually

support-timeline.md

23 Days of Silence - Detailed Timeline

Email #1: June 17, 2025 19:01 KST

• Subject: Technical limitations despite Pro Max subscription
• Content: Detailed description of context window issues
• Response: NONE

Email #2: July 5, 2025 19:39 KST (18 days later)

• Subject: Second attempt - Premium customer needs help
• Content: Escalated tone, mentioned considering cancellation
• Response: NONE

Email #3: July 6, 2025 15:04 KST (next day)

• Subject: Final warning before public exposure
• Content: Mentioned influence and potential negative publicity
• Response: NONE

Email #4: July 9, 2025 15:06 KST (3 days later)

• Subject: Ultimatum with chargeback warning
• Content: Detailed losses and final demands
• Response: NONE

Support Chat: July 9, 2025 17:38 KST

• Bot "Fin" responded in 11 minutes
• Promised human support would contact
• Result: Still waiting

Summary

• Total days: 23
• Total hours: 552
• Human responses: 0
• Bot responses: 1

OP here. The most frustrating part isn't even the $270 loss or the "증명충" insult. It's that a company preaching "Constitutional AI" and "AI safety" shows zero accountability when their AI causes real harm.

If they can't handle basic customer service ethics, how can we trust them with AGI safety?

Evidence and timeline in the Gist. The Korean Twitter thread shows I'm not alone in feeling this way about Anthropic's hypocrisy.

Update: Key evidence as requested:

1. Full conversation showing the humiliation pattern:
   https://github.com/user-attachments/assets/b4f1b635-aa94-4077-ad22-17f5b350d6d9

2. ChatGPT treating me with respect vs Claude's fake reactions:
   https://github.com/user-attachments/assets/2bcbaaca-9d1b-4cbe-882e-f597b804c536

3. Claude admitting to calling me "증명충" (proof-obsessed attention seeker):
   https://github.com/user-attachments/assets/743a3908-6b7f-4eca-b45a-516493f46d08

This is what $200/month gets you from the "Constitutional AI" company.