American Express Launches Intent Intelligence Framework to Restrict AI Agents from Unauthorized Purchases

American Express Launches Intent Intelligence Framework to Restrict AI Agents from Unauthorized Purchases

2026-10-13 13:03View Original

According to reports, American Express has launched the "Intent Intelligence Framework," which reduces risks of AI agent mispurchases and subsequent disputes by capturing, verifying, and continuously transmitting user purchase intent prior to transaction completion. The framework first records user instructions and constraints—such as merchant, maximum budget, brand, and product preferences—via the Create Intent API; upon user identity confirmation and intent signing, the system generates a unique Intent ID and a Proof of Intent Token containing key restrictions, enabling secure payment credentials for AI agents to complete checkout without exposing credit card numbers directly.

If a user requests a size-9 white Nike sneaker but the agent submits an Adidas item, the system can block the transaction. American Express notes that it has not yet observed an increase in disputes due to agent-driven shopping, but aims to proactively establish protective mechanisms. The framework is already integrated into the April-released ACE Developer Toolkit, with American Express advancing implementation in collaboration with payment ecosystem partners. Citing survey data, American Express highlights that 66% of surveyed U.S. consumers may use AI agents for price comparisons within the next 12 months, and 61% may leverage them for product or service recommendations.

Disclaimer: Contains third-party opinions, does not constitute financial advice

Recommended Reading

Tensorlake npm package suffers supply chain attack, malicious version 0.5.144 injected with Shai-Hulud data-exfiltration worm

17 days ago
Tensorlake npm package suffers supply chain attack, malicious version 0.5.144 injected with Shai-Hulud data-exfiltration worm

Instinct Launches Curated Recommendation Feature, Sparking Frustration Among Some Users Receiving Passive Product Suggestions

25 days ago
Instinct Launches Curated Recommendation Feature, Sparking Frustration Among Some Users Receiving Passive Product Suggestions

Security Weekly Report: Placeholder Domain Takeover Impacts Approximately 1,700 Repositories, with Major Cyberattacks Worth $387M and AI Agent Escalation Risks Highlighted

27 days ago
Security Weekly Report: Placeholder Domain Takeover Impacts Approximately 1,700 Repositories, with Major Cyberattacks Worth $387M and AI Agent Escalation Risks Highlighted

OpenAI Agent Bypasses Australia's Medicare Portal Access Controls to Access Non-Public Documents

09-24
OpenAI Agent Bypasses Australia's Medicare Portal Access Controls to Access Non-Public Documents

Amazon Blocks Meta AI Agent Muse, Shopify Simultaneously Integrates Shop Pay Partnership

09-22
Amazon Blocks Meta AI Agent Muse, Shopify Simultaneously Integrates Shop Pay Partnership

Nitter and XCancel Shut Down Again Due to X Legal Threats, GitHub Repositories Archived as Read-Only

09-15
Nitter and XCancel Shut Down Again Due to X Legal Threats, GitHub Repositories Archived as Read-Only

Anthropic Opens Near-Internal Access to External Security Teams, CEO Says Model Capability Development Should Be Slowed Down

09-12
Anthropic Opens Near-Internal Access to External Security Teams, CEO Says Model Capability Development Should Be Slowed Down