Bank of New York Mellon (BNY) is aggressively integrating artificial intelligence across its operations, viewing it as a foundational element of future technology development. The bank’s commitment extends beyond simply adopting AI; it’s building an ecosystem to foster innovation in the space, as evidenced by the recent graduation of five startups from its 2025 Ascent Program.
Sarthak Pattanaik, chief data and AI officer at BNY Mellon, articulated the bank’s perspective: “We think AI will be the chassis of any technology product built going forward. It’s like part of the operating system. So, for us, [it’s] all…” – a statement cut short in the provided source material, but indicative of the central role AI is expected to play.
The Ascent Program, led by BNY’s Strategic Partnerships, Investments & Innovation (SPIN) team, is a six-month proof-of-concept (POC) platform designed to validate solutions from early-stage companies within real-world financial services environments. This year’s cohort focused on challenges spanning AI governance, cyber defense, market intelligence, and operational risk. The program provides startups with access to BNY’s leadership and rigorous testing frameworks, accelerating product development and market readiness.
The five graduating companies represent a diverse range of AI applications critical to the evolving financial landscape. Protect AI, poised for acquisition by Palo Alto Networks, offers real-time AI security, including vulnerability scanning and threat prevention. Calypso AI focuses on securing the entire AI application lifecycle through AI red teaming and defensive security solutions. Reality Defender addresses the growing threat of deepfakes with an API-driven platform for identifying AI-generated impersonation fraud. AgentSmyth delivers real-time investment intelligence to institutional traders via AI-powered decision support, and Deep Tempo specializes in predictive cybersecurity, utilizing deep learning to proactively detect and prevent cyber threats.
The emphasis on cybersecurity within the Ascent Program highlights the increasing importance of protecting AI systems themselves. As financial institutions increasingly rely on AI for critical functions, the potential for malicious attacks targeting these systems grows. Protect AI and Deep Tempo’s solutions directly address this concern, offering proactive defenses against vulnerabilities and threats.
Beyond cybersecurity, the cohort also demonstrates the potential of AI to enhance core financial services. AgentSmyth’s investment intelligence platform exemplifies how AI can provide a competitive edge to institutional traders, while Reality Defender’s deepfake detection capabilities are crucial for mitigating fraud risks in an increasingly digital world. The inclusion of Calypso AI underscores the need for robust security measures throughout the entire AI development process, from initial design to deployment and ongoing maintenance.
BNY Mellon’s investment in AI extends beyond the Ascent Program. In May 2025, James Cunningham, Senior Director at BNY Mellon, participated in a webinar discussing the optimization of AI and cloud transformation to modernize financial services. The discussion centered on navigating the dynamic regulatory landscape surrounding these technologies, balancing innovation with compliance, and addressing the risks and opportunities presented by blockchain and distributed-ledger technology (DLT).
The webinar, also featuring representatives from Nasdaq, Redcliffe Capital, and Amazon Web Services, highlighted key discussion points including mitigating risks while leveraging AI and cloud, understanding evolving regulatory expectations, and addressing the impact of the T+1 settlement cycle on financial infrastructure. This participation signals BNY Mellon’s proactive engagement in shaping the future of financial technology and its commitment to responsible innovation.
However, the broader adoption of AI in financial risk management isn’t without its challenges. According to industry analysis, a primary obstacle is gaining access to high-quality data and ensuring proper integration with existing systems. This underscores the need for financial institutions to invest in data infrastructure and develop robust data governance frameworks to support AI initiatives.
The cautious approach to embracing large language models (LLMs) within the financial sector, as noted in February 2024, further illustrates the complexities of AI implementation. While LLMs offer significant potential for enhancing efficiency and decision-making, concerns remain regarding their accuracy, reliability, and potential for bias. Christian Hull, former head of innovation, EMEA, at BNY Mellon, acknowledged the need for careful consideration and fine-tuning of LLMs to ensure they are appropriate for use in financial applications.
Alexander Sokol, Founder and Executive Chairman of CompatibL, emphasized that current LLMs often struggle with the specialized vocabulary and complex concepts prevalent in the financial industry, requiring human supervision even for tasks like generating regulatory filings. This highlights the importance of tailoring AI models to specific financial use cases and investing in ongoing monitoring and validation.
Looking ahead to 2026, financial system modernization will continue to be a key theme, with AI playing a central role. BNY Mellon’s initiatives, including the Ascent Program and its participation in industry discussions, position the bank as a leader in this transformation. The success of the program’s graduates and the ongoing exploration of AI applications suggest a future where AI is deeply integrated into the fabric of financial services, driving innovation, enhancing security, and improving efficiency.
