Broadridge Joins Anthropic’s Project Glasswing

Broadridge Financial Solutions announced it has joined Anthropic’s Project Glasswing to evaluate the use of Anthropic’s large language models in Broadridge’s operations. The initial phase will consist of pilots and proofs of concept focused on business workflows.

Broadridge, which provides investor communications and technology services to banks, broker-dealers and asset managers, said the engagement will assess model capabilities, safety controls and integration approaches prior to any broader deployment.

Under Project Glasswing, Anthropic supplies partners with access to its models plus tools and processes designed to manage risk and compliance in enterprise settings. Broadridge will design controlled tests intended to preserve client data privacy and meet regulatory requirements relevant to financial services.

Planned tests include customer-facing communications, document analysis and automation of routine back-office tasks. Work will take place within Broadridge’s secure environments and follow stepwise evaluation protocols.

Broadridge will evaluate how model outputs can be audited, how data flows can be isolated and which guardrails are required to align model behavior with legal and compliance standards. Any adoption beyond the pilot stage will require validation against the company’s internal controls and industry rules.

Anthropic develops large language models and pairs them with safety engineering and operational practices aimed at reducing risk when models are used on sensitive tasks. The agreement follows a broader trend of financial firms conducting controlled tests with model providers to determine where generative AI can be applied without compromising compliance or client confidentiality.