- Building AI on credit: structures and spilloversFactual summary
The BIS FSI published research on the growing use of long-term debt and off-balance sheet structures to finance AI infrastructure.
So whatComplex leverage and off-balance sheet debt for AI data centres introduce systemic counterparty and credit risks for financial institutions.
Do whatReview credit risk exposures and indirect leverage associated with AI infrastructure and debt-financed data centre developments.
- Bank of America kindly asks that you don’t share your banking password with your AI assistantFactual summary
Bank of America warned customers against sharing banking passwords with AI agents, noting users remain liable for resulting activity.
- Amex Publishes Intent-Authentication Framework for AI Agent PaymentsFactual summary
American Express published a framework to capture and authenticate customer instructions to AI agents, shifting payment liability to intent records.
So whatEstablishing intent-based authentication models for autonomous transactions shifts payment liability and creates an industry precedent for agentic payment workflows.
Do whatReview agentic payment authentication protocols and liability frameworks with the team responsible for payment systems technology.
- Backbase Helps Banks Deploy Agentic AI With Delivery FactoryFactual summary
Backbase launched Delivery Factory, offering AI agents to assist bank engineering teams from requirements to code generation.
- OpenAI and Anthropic Want the Junior Banker’s 2 a.m. ShiftFactual summary
OpenAI, Anthropic, and financial data incumbents are competing to automate junior investment banking research and analysis workflows.
- UK FS sector embraces GenAIFactual summary
A joint BoE and FCA survey reveals generative AI now accounts for over half of all AI systems used in UK financial services.
So whatRegulatory survey data shows rapid GenAI adoption across UK finance, accelerating supervisory scrutiny on model governance and operational resilience.
Do whatBenchmark your internal AI deployment inventory and governance frameworks against official UK regulatory survey data.
- Bank of America, Blackstone, Citi and Morgan Stanley Announce Landmark $63 Billion Financing Structure to Enable AI Compute Capacity in Partnership with BroadcomFactual summary
Bank of America, Blackstone, Citi, and Morgan Stanley announced a $63B financing structure to expand AI compute for Anthropic via Broadcom.
So whatMajor financial institutions are structuring novel, large-scale capital solutions to directly back the hardware and compute infrastructure required for enterprise AI models.
Do whatReview capital exposure and structured finance opportunities related to mega-scale AI infrastructure compute projects with investment and risk teams.
- Cross-request KV cache exposure in the GLM-5.3 Workers AI model, and our remediation with SGLangFactual summary
Cloudflare resolved a Workers AI configuration error that triggered an SGLang bug, exposing cross-request KV cache data in GLM-5.3.
So whatMulti-tenant LLM inference infrastructure carries subtle memory-isolation risks that can leak sensitive prompts across distinct sessions.
Do whatReview vendor data-isolation guarantees with the team responsible for enterprise AI platform operations before using shared inference endpoints.
- Pretraining data filtering in open-weight biological modelsFactual summary
The UK AI Security Institute finds pretraining data filtering for open-weight protein models effective but vulnerable to fine-tuning.
So whatData filtering safeguards on open-weight models degrade under subsequent fine-tuning, complicating risk management.
Do whatMonitor findings on open-weight model vulnerabilities through the team responsible for model risk management.
- A Survey on the Security of Long-Term Memory in LLM Agents: Toward Mnemonic SovereigntyFactual summary
A research survey identifies emerging security risks in LLM agents with persistent, long-term memory, including cross-session poisoning and unauthorized access.