AI x Fintech Weekly — 2026-06-25

Signals this week

• Agents & tooling (3) — Mastercard and PrivatBank complete Ukraine’s first payment executed by an AI agent; Neglected Free Lunch from Post-training: Progress Advantage for LLM Agents; GLM-5.2 is a step change for open agents
• Payments & rails (3) — Mastercard and PrivatBank complete Ukraine’s first payment executed by an AI agent; Newland brings Android payments terminal to Poland; Iceland to join Target Instant Payment Settlement System
• Risk & compliance (1) — Privacy Vulnerabilities of Attention Layers in Tabular Foundation Models and Protection of High-Risk Queries

Top links (curated)

• Mastercard and PrivatBank complete Ukraine’s first payment executed by an AI agent (Finextra) — https://www.finextra.com/pressarticle/110244/mastercard-and-privatbank-complete-ukraines-first-payment-executed-by-an-ai-agent
• Privacy Vulnerabilities of Attention Layers in Tabular Foundation Models and Protection of High-Risk Queries (arXiv cs.AI) — https://arxiv.org/abs/2606.26021v1
• Neglected Free Lunch from Post-training: Progress Advantage for LLM Agents (arXiv cs.AI) — https://arxiv.org/abs/2606.26080v1
• TriViewBench: Controlled Complexity Scaling for Multi-View Structural Reasoning in MLLMs (arXiv cs.AI) — https://arxiv.org/abs/2606.26029v1
• AI, HR and financial services: Modernising banking systems from front to back (Finextra) — https://www.finextra.com/event-info/619/ai-hr-and-financial-services-modernising-banking-systems-from-front-to-back
• Digital sovereignty: The key deliverable of banking transformation (Finextra) — https://www.finextra.com/event-info/616/digital-sovereignty-the-key-deliverable-of-banking-transformation
• Newland brings Android payments terminal to Poland (Finextra) — https://www.finextra.com/pressarticle/110248/newland-brings-android-payments-terminal-to-poland
• Iceland to join Target Instant Payment Settlement System (Finextra) — https://www.finextra.com/newsarticle/47983/iceland-to-join-target-instant-payment-settlement-system
• GLM-5.2 is a step change for open agents (Hacker News) — https://www.interconnects.ai/p/glm-52-is-the-step-change-for-open
• Anthropic says Alibaba illicitly extracted Claude AI model capabilities (Hacker News) — https://www.reuters.com/world/china/anthropic-says-alibaba-illicitly-extracted-claude-ai-model-capabilities-2026-06-24/