AI in financial services, week ending 16 October 2026

Week ending 16 October 2026 · 44 sources

Banks shut the door on outside AI agents as supervisors tighten

JPMorgan Chase and Schwab moved to block unauthorised outside AI agents from customer accounts, and Jamie Dimon warned customers of a 'liability transfer' 123. The FSB finalised 12 sound practices for AI adoption and several supervisors flagged AI cyber risk 456.

Which way the market is moving

Banks gate outside AI agents

JPMorgan Chase and Schwab are blocking unauthorised outside AI agents from customer accounts, putting agent liability on the table 123.

JPMorgan Chase is blocking outside AI agents, including Muse and Instinct, from directly accessing customer accounts 117. Barron's reported that Schwab has also blocked unauthorised agents 2. Jamie Dimon warned customers of a 'liability transfer' when they hand tasks to AI agents and urged them to understand the risks 318.

Not every bank is taking the same line. American Banker reported that some banks are exploring 'bring your own AI agent' models, while bankers grapple with liability when agents make mistakes 87. Agent demand is growing: Stripe's John Collison said its agentic commerce activity went 'vertical' last month 19.

What it means: Institutions need a clear position on third-party agent access, covering authentication, permitted actions and who bears the loss when an agent errs. Customer terms and fraud controls may need review before agent traffic grows further.

Questions for vendors:

  1. How does your platform tell an authorised customer agent apart from a scraper or rogue bot?
  2. Do you support scoped, delegated credentials for agents, and can they be revoked in real time?
  3. How are agent-initiated actions logged so disputes and liability can be resolved?

In the registry: AgentNext (BUSINESSNEXT), OCTO (TCG Process), Insurer Solution (Acturis)

Supervisors tighten AI governance and cyber expectations

The FSB finalised 12 sound practices for AI adoption while several supervisors focused on AI-enabled cyber risk 4569.

The Financial Stability Board finalised 12 sound practices to guide AI adoption by banks and other financial institutions 4. PYMNTS reported the FSB wants firms to govern AI across the whole company 20. US and European regulators are also stepping up scrutiny of AI safeguards, Chinese models and cybersecurity 2122.

Cyber risk from AI is a common thread. The Bank of England warned banks about AI cybersecurity risks 5, and New Zealand's RBNZ and FMA set out expectations on cyber risks from frontier AI 6. In Korea, AI-driven attacks hit 16 financial firms and exposed 100,000 records, and the FSC ordered bank-wide checks 239.

What it means: AI governance is now expected to be enterprise-wide and evidenced, not run project by project. Cyber teams should assume supervisors will ask how AI changes both attack exposure and recovery plans.

Questions for vendors:

  1. Which of the FSB sound practices does your product help us evidence, and how?
  2. Where are your models built and hosted, and do any components come from jurisdictions regulators are asking about?
  3. How do you test your AI features against AI-assisted attacks and prompt injection?

In the registry: AgentNext (BUSINESSNEXT), Insurer Solution (Acturis), OCTO (TCG Process)

Vendors build agents into compliance, core and finance

Vendors launched agents for compliance, sanctions screening, core banking and business finance work 24252627.

FIS launched an AI banking agent, which Forbes framed against a trust test for fintech compliance 24. GSS partnered with Storm Reply on an AI-powered cloud capability for sanctions screening 25. Customers Bank and Palantir are collaborating on an AI-native future for banking 28, and Emprise Bank signed a design partnership with Maximum to transform its core system 26.

Agents are also reaching business customers. PEX launched PEX AI to run multi-step finance workflows and connect to assistants such as ChatGPT and Claude 27, and Extend offered an AI Receipt Agent to bank partners' small business customers 29. Citi's Jane Fraser pointed to reduced vendor spend, while Goldman's David Solomon urged 'rigorous testing' and 'clear human accountability' 30.

What it means: Agentic features increasingly arrive inside existing vendor relationships. Procurement should treat each as a new AI use case needing testing, named human accountability and exit terms.

Questions for vendors:

  1. Which decisions can your agent take without human approval, and can we configure those limits?
  2. What testing results can you share for the agent on data like ours?
  3. How do we export agent outputs and audit trails if we leave your platform?

In the registry: AgentNext (BUSINESSNEXT), Post-trade operations (Xceptor), OCTO (TCG Process)

Sovereign and owned compute gains ground

Starling, SBS, Farsight and Deutsche Bank moved AI onto sovereign clouds or institution-controlled compute 11131231.

Starling is working with Nvidia to build a transactional foundation model trained and hosted within a UK sovereign cloud 11. SBS launched a sovereign AI platform for core banking with NVIDIA 3213. Farsight partnered with Ornn to help banks run AI on their own compute 12, and brought secure AI workflows to Deutsche Bank 31.

Model provenance is becoming a selection test. Napier AI said institutions choosing AI for financial crime compliance now weigh where a model was trained, who built it and what data it learned from 33. NatWest used Nvidia speech recognition so its audio-visual spending insights tool understands regional accents 34.

What it means: Data residency and model provenance are moving from contract clauses to architecture choices. Institutions should know which AI workloads must stay in jurisdiction and which can use shared models.

Questions for vendors:

  1. Can the model be trained and run within our chosen jurisdiction or on our own infrastructure?
  2. What data was the model trained on, and can you document its provenance?
  3. How do cost and performance differ between your hosted and on-premise options?

In the registry: RadiantOne Platform (Radiant Logic), Vibe (Mistral AI), Loan Origination (timveroOS)

AI risk becomes an insurance and capital question

Insurance emerged as an obstacle and a funding source for AI, while insurers scaled their own platforms 14151635.

The Wall Street Journal reported that insurance is AI's next big obstacle 14. Harbor emerged from stealth with $10M in funding to handle AI-agent risk 15. On the capital side, CoreWeave is targeting what it calls 'unlimited' insurance money to fund its AI build-out 163637.

Insurers are also scaling their own use. Manulife launched an enterprise AI platform and is targeting $1 billion in value by 2027 35. A KPMG AI survey gave New Zealand brokers questions to put to insurers 38, and ERGO NEXT rolled out an AI property inspection tool to speed up commercial property cover 39.

What it means: Insurers meet AI as underwriters of AI liability, as investors in AI infrastructure and as users. Other institutions should check whether existing cyber and liability cover responds to errors made by AI agents.

Questions for vendors:

  1. Does your contract allocate liability for errors made by your AI agents, and is that backed by insurance?
  2. What testing and monitoring evidence can you provide to support our insurer's underwriting?

In the registry: AgentNext (BUSINESSNEXT), Case Management (SymphonyAI), OCTO (TCG Process)

Regulation and supervision

Sources

  1. CNBC, 2026-10-13: JPMorgan Chase blocks outside AI agents like Muse, Instinct from directly accessing accounts
  2. Barron's, 2026-10-15: JPMorgan, Schwab Block Unauthorized AI Agents. Why That Matters to Customer Cash.
  3. Banking Dive, 2026-10-13: JPMorgan CEO Dimon warns customers of ‘liability transfer’ with AI
  4. ABA Banking Journal, 2026-10-15: Financial Stability Board finalizes ‘sound practices’ for AI adoption
  5. PYMNTS.com, 2026-10-15: BoE Warns Banks Against AI Cybersecurity Risks
  6. Finextra (security), 2026-10-15: New Zealand regulators set out cyber and AI risk expectations
  7. American Banker, 2026-10-13: Bankers grapple with liability when AI agents mess up
  8. American Banker, 2026-10-15: Why some banks are exploring 'bring your own AI agent'
  9. FinTech Global, 2026-10-13: Korea’s FSC orders bank-wide checks after AI cyberattacks
  10. The Japan Times, 2026-10-15: Banking watchdogs scrutinize AI use amid stability fears, sources say
  11. Finextra, 2026-10-14: Starling becomes first bank to train AI model on UK sovereign cloud
  12. Yahoo Finance, 2026-10-15: Farsight Partners With Ornn to Help Banks Run AI on Their Own Compute
  13. Yahoo Finance, 2026-10-14: SBS launches sovereign AI platform for core banking with NVIDIA
  14. WSJ, 2026-10-16: AI Runs Into Its Next Big Obstacle: Insurance
  15. Insurance Journal, 2026-10-13: Harbor Emerges From Stealth with $10M Funding to Handle AI-Agent Risk
  16. Financial Times, 2026-10-14: CoreWeave targets ‘unlimited’ insurance money to fund AI build-out
  17. Finextra Research, 2026-10-15: JPMorgan Chase blocks outside AI agents from directly accessing customer accounts
  18. Yahoo Finance, 2026-10-13: JPMorgan’s Jamie Dimon Warns Of Rogue AI Agents, Urges Customers To Understand Risks
  19. Payments Dive, 2026-10-15: Stripe agentic commerce revs up
  20. PYMNTS.com, 2026-10-15: FSB Says Banks Need to Have Company-Wide AI Regulation
  21. Yahoo Finance, 2026-10-15: AI Agents Could Pose New Risks to Banks: US, European Regulators Step Up Scrutiny of AI Safeguards, Chinese Models and Cybersecurity: Report
  22. Yahoo Finance, 2026-10-13: Market Chatter: US Regulators Press Banks on AI Protections
  23. The Korea Times, 2026-10-14: AI-driven cyberattacks hit 16 financial firms, expose 100,000 records
  24. Forbes, 2026-10-15: FIS Launches AI Banking Agent As Fintech Compliance Faces A Trust Test
  25. Finextra (cloud), 2026-10-14: GSS and Storm Reply collaborate on sanctions screening
  26. Finextra, 2026-10-15: AI-native core banking startup Maximum signs Emprise Bank
  27. Finextra, 2026-10-15: PEX brings AI-powered finance workflows to ChatGPT and Claude
  28. Yahoo Finance, 2026-10-15: Customers Bank and Palantir Collaborate to Build an AI-Native Future for Banking
  29. Finextra, 2026-10-15: Extend brings AI receipt agent to bank partners' SMB customers
  30. Banking Dive, 2026-10-15: What Wells Fargo, Citi and Goldman Sachs’ CEOs are saying about AI
  31. FinTech Global, 2026-10-13: Farsight brings secure AI workflows to Deutsche Bank
  32. The Paypers, 2026-10-14: SBS launches sovereign AI platform for banks with NVIDIA
  33. FinTech Global, 2026-10-15: Sovereign AI vs shared intelligence: AML’s new dilemma
  34. Finextra, 2026-10-15: NatWest enlists Nvidia to help its AI spending tool understand regional accents
  35. Insurance Business, 2026-10-14: Manulife launches enterprise AI platform as it targets $1 billion in value by 2027
  36. Startup Fortune, 2026-10-14: CoreWeave wants insurers to help bankroll its AI data center boom
  37. Seeking Alpha, 2026-10-14: CoreWeave seeks insurers’ capital to finance AI infrastructure build-out - report (CRWV:NASDAQ)
  38. Insurance Business, 2026-10-14: KPMG AI survey gives NZ brokers questions to put to insurers
  39. FinTech Global, 2026-10-15: ERGO NEXT unveils AI tool for 15-minute property inspections
  40. The Australian, 2026-10-13: Fears AI harm is near for banks. Watchdogs step up containment efforts.
  41. Bloomberg.com, 2026-10-13: More Weak Banks in IMF Doomsday Scenario as AI Boom Fuels Risks
  42. PYMNTS.com, 2026-10-13: Banking Regulators Tighten Oversight of Possible AI Risks
  43. ECB, 2026-10-13: Philip R. Lane: The AI opportunity for Europe
  44. Finextra (cloud), 2026-10-14: DOCOsoft contracted by Siriuspoint for claims management

About AI software

AI software across financial services: AI-native products and the machine learning, generative AI and agent capabilities inside banking, insurance and investment software.

Browse the products in the registry

Get the market overview every Monday: subscribe to the Vederca newsletter.