BankingNewsAI Daily Brief ·
OpenAI introduces the Agents API
Banking AI
Financial institutions & fintech technology
OpenAI introduces ChatGPT for Financial Services
For the CIO of a regional bank
A packaged workspace does not remove your control burden; it moves premium data and analysis into one service. Do not let analysts use it until entitlement rules, citations and required human review work in the same workflow.
OpenAI introduced ChatGPT for Financial Services, a tailored ChatGPT Work experience for research, financial models and client materials. It combines GPT-6 Astra with data from Daloopa, PitchBook and LSEG News, which OpenAI indexes and hosts on its infrastructure. The product provides granular citations so users can trace figures and claims to sources while they analyze. OpenAI is also working with S&P Capital IQ, LSEG, MSCI, Dow Jones Factiva and Moody’s on shared sign-in and entitlement integrations for existing subscribers. Administrators can publish Excel, Word and PowerPoint templates for valuation models, research notes and pitchbooks.
→ Action
Data governance: Test provider entitlements, source citations and required reviewer approval before enabling hosted financial-data workflows.
Read article → from OpenAI
DFS issues cybersecurity risk-assessment guidance
For the CISO at a DFS-regulated bank
Treat AI rollouts, critical-system changes and cloud concentration as linked technology decisions. Each needs a documented risk-assessment trigger and a clear control or risk-acceptance response.
New York DFS issued guidance on risk assessments for regulated financial services entities’ cybersecurity programs. Its cybersecurity regulation requires entities to review and update assessments at least annually and when a business or technology change materially changes cybersecurity risk. The guidance says entities should reassess before or after a major system migration, acquisition, or new critical-system implementation. It also calls for assessment of common cloud, managed-service and software dependencies, and of AI-related threat exposure, data risks, access controls and third-party dependencies. DFS says the guidance adds no new obligations but clarifies existing requirements on governance, methodology, scope, documentation and risk-informed controls.
→ Action
Cybersecurity risk: Map AI deployments, critical-system changes and common third-party dependencies to documented risk-assessment triggers and control-update decisions.
Read article → from Ny
Experian deploys real-time AI decisioning across lending marketplaces
For the consumer lending head at a regional bank
Experian can route applicants using data and rules that sit outside your own funnel. Decide whether its lead quality offsets the work to validate underwriting fit, pricing, fair-lending outcomes and acquisition cost.
Experian added Experian Activate, an AI-enabled decisioning engine, to Experian Marketplace for credit cards, personal loans and car insurance. The engine combines credit data, analytics, AI, consumer-permissioned financial information and lender-specific underwriting criteria. It uses real-time credit intelligence to provide relevant offers and includes permissioned cash-flow data to give lenders a more complete view of financial health. Experian says the cash-flow data helps it reach qualified customers more efficiently. Experian also powers third-party experiences including ChatGPT, where it added insurance in February, personal loans in June and credit cards in August.
→ Action
Fair lending: Test approval and pricing outcomes for Activate leads against existing acquisition channels.
Read article → from PYMNTS
General AI
Large language models & AI infrastructure
OpenAI introduces the Agents API
For the technology-risk leader at a US bank
OpenAI makes long-running agents easier to deploy without owning the harness, but your bank still owns the risk. Set location, credential, data-retention and action-approval rules before agents get access to bank tools.
OpenAI launched its Agents API in public beta, offering developers a managed Codex harness and infrastructure for cloud agents. Developers can choose an OpenAI-managed sandbox, their own infrastructure, or sandbox partners including Cloudflare, Oracle and Vercel. Agents can run for days, work with files, run code and save intermediate results; OpenAI hosts and maintains the harness. The API supports MCP, custom functions, web search, tool search, parallel tool calls and subagents, while automatically compacting earlier context as sessions near their context limit. Developers pay for tokens and tools, with no added Agents API fee during beta.
→ Action
Technology risk: Map sandbox location, secret access, file retention and approval gates before any agent receives bank tools or data.
Read article → from OpenAI
Salesforce completes acquisition of Fin
For the customer-service head at a bank
Salesforce now owns a customer-agent platform that could enter your service stack. Do not treat it as a simple add-on: require proof of answer accuracy, human escalation paths and customer-data controls before deployment.
Salesforce completed its acquisition of Fin, formerly Intercom. Fin’s customer agent platform joins Salesforce. Salesforce said the platform gives companies faster, more flexible ways to automate customer service. Salesforce said it can deliver measurable outcomes.
→ Action
Customer service: Test answer accuracy, human-escalation rules and customer-data handling before enabling Fin for customer interactions.
Read article → from Salesforce
Get this in your inbox every morning
Free · No spam · Unsubscribe anytime