· Transaction & Payment Processing Services · S&P 500
PYPL PAYPAL HOLDINGS INC $44.7B
No. Fraud prevention is a list in a risk factor. No program, no result.
5 of 15
Reach, Weight, and Proof.
Last updated score 22 OctLast review 29 OctLast earnings call 27 OctNext earnings call not posted yet
What they are doing
PayPal's 2025 annual report says it uses artificial intelligence and machine learning in fraud prevention, risk management, product optimization, and customer service. That sentence sits in a regulation risk factor. No program is named, and no result is given. The same filing says PayPal wants shoppers to be able to pay through AI agents. That is a checkout surface, not a score inside authorization or fraud. The reported transaction-loss rate is not tied to a model. On the 27 October 2026 call, Enrique Lores said a new risk platform, using real-time data and AI, cut credit decision times nearly in half in Canada, Italy, and Spain. He did not tie that platform to the transaction-loss rate. AI chat agents resolved 24 million contacts. That is a help line. Agentic checkout is not a fraud score. The published score is still 5. The recommendation, not yet accepted, is Reach 2, Weight 2, Proof 3, total 7.
Why the score
- Reach 1 not 2: the uses are a list, not one function we can see running.
- Weight 2 not 3: named fraud prevention, with no project and no result. Agent checkout does not change the payment decision.
- Proof 2 not 3: named in the 10-K and tied to fraud prevention. Not in the loop.
What would move it next quarter: A fraud or approval rate on the risk platform. Decision time in three countries will not move Weight to 3.
Open these first
Evidence
PayPal says it uses AI and machine learning in fraud prevention, risk management, product optimization, and customer service. The sentence is in the artificial-intelligence regulation risk factor. No program and no result are named.