- FIS AML Investigations Agent Now Commercially Available to Help Banks Tackle the Scale and Cost of Financial Crime ComplianceFactual summary
FIS makes its Anthropic-powered AML Investigations Agent commercially available to all US banks.
- JPMorgan Chase blocks outside AI agents from directly accessing customer accountsFactual summary
JPMorgan Chase CEO Jamie Dimon stated the bank will block external AI agents from directly accessing retail customer accounts.
So whatDirect account access by external consumer AI agents creates severe security and liability risks that tier-one banks are actively moving to block.
Do whatReview open-banking permissions and API authentication controls with the cybersecurity team to ensure third-party agents cannot directly scrape or access account data.
- Jamie Dimon Slams the Door on Rogue AI AgentsFactual summary
JPMorgan Chase CEO Jamie Dimon stated the bank will block unvetted external AI agents from accessing its enterprise systems.
So whatMajor G-SIBs are restricting external agentic access, setting a firm perimeter security benchmark for financial institution data integration.
Do whatReview external API access controls and agent authentication policies with your information security team.
- Starling becomes first bank to train AI model on UK sovereign cloudFactual summary
Starling Bank is collaborating with Nvidia to train and host a transactional foundational model within a UK sovereign cloud.
- Stripe agentic commerce revs upFactual summary
Stripe President John Collison reported that the payments firm's agentic commerce activity surged rapidly last month.
- 5 Things JPMorgan’s AI Agent Warning Means for BankingFactual summary
JPMorgan executives warned of AI agent risks, specifically highlighting security threats from credential sharing and screen scraping.
- Cohere to acquire AISLE to advance cyber-resilient sovereign AIFactual summary
Cohere has announced its intention to acquire AISLE to enhance cyber-resilient sovereign AI deployment options for enterprise customers.
- Build long-horizon agents with Grok 4.7 on Amazon BedrockFactual summary
AWS made Grok 4.7 available on Amazon Bedrock, publishing benchmarks on tool-calling performance and token costs across reasoning levels.
- Prompt Injection Beats Frontier Models About Half the Time, and Audio Models Mostly Ignore the AudioFactual summary
Recent research syntheses indicate prompt injection attacks succeed roughly 50% of the time against frontier AI models.
So whatHigh prompt injection success rates signal that enterprise deployments involving autonomous actions require strict security controls around untrusted inputs.
Do whatReview prompt injection vulnerabilities with the team responsible for model security controls before deploying autonomous agents.
- Ideological Bias in LLMs' Economic Causal ReasoningFactual summary
Research finds LLMs exhibit systematic ideological bias in economic causal reasoning, particularly on policy-contested topics.