BankingNewsAI Daily Brief ·
CISA deploys Anthropic’s Mythos model to audit federal software for security vulnerabilities.
Banking AI
Financial institutions & fintech technology
Valley Bank is using tiered employee access to put cost controls around enterprise AI
Valley Bank’s COO says the lender is managing AI economics through tiered employee access while using the technology for internal efficiency and relationship-building. The practical shift is from a simple build-versus-buy decision to an operating-model question: which users get which capabilities, at what cost and with what measurable return.
Action
Adopt role-based AI entitlements and unit-level usage reporting before broad licenses turn experimentation into an unmanaged recurring cost.
Tangos AI raised $20 million to automate financial-crime and compliance investigations
Israeli fintech Tangos AI has raised a $20 million seed round for autonomous AI tools focused on financial-crime, fraud and compliance investigations. This is a targeted bet on agentic workflows in a heavily staffed, evidence-intensive bank function—not a generic banking copilot.
Action
Pressure-test investigation workflows for where agents can assemble evidence, draft cases and route exceptions while preserving investigator review and audit trails.
General AI
Large language models & AI infrastructure
xAI’s Grok 4.5 enters the frontier coding-agent market at a materially lower price point
xAI launched Grok 4.5, its first model trained specifically for coding and agents, through its API, Grok Build, Cursor and other agent platforms. It is priced at $2 per million input tokens and $6 per million output tokens; Artificial Analysis ranked it fourth on its broad intelligence and agent evaluations and found coding-agent performance comparable to GPT-5.5 in Codex with substantially lower token use. The release adds a credible lower-cost frontier-model option to the enterprise AI vendor market dominated by OpenAI and Anthropic.
Action
Benchmark Grok 4.5 against your current coding-agent and operations-automation models, especially for high-volume workflows where token cost and latency drive the business case.
CISA is using Anthropic’s Mythos model to audit federal software for vulnerabilities
The U.S. Cybersecurity and Infrastructure Security Agency is using Anthropic’s Mythos AI to audit government software for vulnerabilities. A national-security deployment is a meaningful validation of AI-assisted code review, while also raising the bar for controls around sensitive code, findings and human remediation decisions.
Action
Expand secure AI-assisted code-review trials in non-production repositories and define escalation, evidence-retention and human sign-off requirements before using findings in critical systems.