AI in Banking news rundown:
1. Robinhood Launches AI Agents for Stock Trading and an Agentic Credit Card
Robinhood on May 27 launched two agentic products: an AI trading agent that can autonomously execute equity trades on a customer's behalf, and a credit card whose spending decisions can be delegated to an AI agent. The move puts Robinhood ahead of every major retail bank in deploying autonomous, customer-facing financial agents at scale. Users set parameters and the agent acts without requiring per-transaction approval. This is the first mainstream US brokerage to hand execution authority - not just recommendations - to an AI model.
2. ING Is Using Vibe Coding AI to Build Its New Trading Systems
ING disclosed on May 26 that it is using so-called vibe coding - where engineers describe intent in natural language and AI generates working code - to build production trading infrastructure. The bank has not named the specific model stack but confirmed the approach is live in development pipelines, not a pilot. ING joins Goldman Sachs and JPMorgan in deploying AI-generated code in core financial systems, but is the first European bank to publicly confirm vibe coding at the trading-system level. The shift compresses delivery cycles that historically ran 12-18 months for comparable infrastructure.
3. BNP Paribas Partners With Mistral to Prepare for Mythos-Class AI Models
BNP Paribas confirmed on May 26 it is working with French AI lab Mistral to build internal infrastructure capable of running frontier reasoning models comparable to Anthropic's Mythos. The partnership is positioned as a sovereignty play - keeping sensitive banking data on European models rather than routing through US hyperscaler APIs. BNP has not disclosed contract value, but the tie-up directly mirrors the Anthropic Mythos story from May 19, when US regulators delayed bank cyberattack tests because Mythos outperformed the red-team scenarios. BNP's move signals that Tier 1 European banks are racing to have model-ready architecture before the next generation of reasoning AI ships.
4. Alipay Launches AI Wallet and Token Pay to Power Agentic Economy
Alipay on May 26 announced AI Wallet and Token Pay, a next-generation payment infrastructure designed explicitly for AI agent-to-agent transactions rather than human-initiated payments. Token Pay allows AI agents to hold, spend, and transfer funds autonomously within permissioned bounds set by the account holder. The infrastructure targets the emerging agentic economy where software agents book services, pay invoices, and settle micro-transactions without human confirmation at each step. Alipay serves over 1.3 billion users and the move positions Ant Group as the default payments rail for agentic commerce in Asia before Western incumbents have published specs.
5. Anthropic Mythos Prompts US Regulators to Delay Bank Cyberattack Tests
US bank regulators postponed scheduled cyberattack simulation exercises in mid-May after Anthropic's Mythos reasoning model demonstrated capabilities that exceeded the assumptions built into existing red-team scenarios, according to Bloomberg reporting on May 19. The delay affects multiple large US banks that were due to run DFAST-adjacent resilience tests. Regulators have not given a rescheduled date and are reportedly consulting with AI labs to rebuild threat models. It is the first documented case of a new AI model forcing a formal pause in US bank supervisory activity.
#news #banking #aibanking #artificialintelligence