PCI SSC calls for human approval of AI agent actions involving cardholder data
PCI SSC guidance urges human approval before AI agents act on cardholder data in payment environments.
The PCI Security Standards Council published Security Considerations for AI Systems, advisory guidance on protecting payment data used by AI and on AI-assisted attacks. It calls for least-agency design, a named human owner of AI output, and explicit human approval before agents act on cleartext cardholder data. Passwords and cryptographic keys should stay out of prompts, context, outputs, and logs, with independent data-loss prevention and scoping that treats decryptable data as readable. The document also covers shadow AI, adversarial testing, prompt injection, model poisoning, and contracts barring providers from training on customer data.
- The guidance is advisory; existing PCI DSS requirements still prevail.
- Agents handling cleartext cardholder data should require human approval.
- Avoid combining sensitive data, external communications, and untrusted input.
- Keep passwords and keys out of prompts, context, outputs, and logs.
- Plans should cover prompt injection, model poisoning, and shadow AI.
- Products
- PCI DSS
- Organizations
- PCI Security Standards Council
From the original article
3 min read at helpnetsecurity.com
The PCI Security Standards Council (PCI SSC) has published Security Considerations for AI Systems, guidance covering the protection of data supplied to AI systems in payment environments and defenses against AI-assisted attacks. Developed with industry stakeholders, the document addresses governance, deployment, access controls, testing and the application of PCI standards.
By Anamarija Pogorelec · helpnetsecurity.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.