Hopefully you heard about our Microsoft collaboration either at Microsoft Ignite or in the Microsoft blog. Let me take a few minutes to put this news into context by explaining the problem we are trying to solve.
One of the biggest concerns about artificial intelligence (AI) from all stakeholders—consumers, regulators, and corporate leaders—is safety. Saifr is on a mission to make AI safer.
AI and language models, both large and small language models (LLMs and SLMs, respectively), can be transformative tools—that’s why the technology is seeing quick, widespread adoption and integration. In many cases, AI technology can automate manual, rote processes, increasing efficiency while providing more robust insights. LLMs and SLMs, which are a subset of generative AI, can be particularly helpful when creating content: a use case that is exploding. This is great news! However...
All technology, especially emerging technology, needs to be safe in how it’s developed, how it’s used, and the output it produces. Consumers and increasingly regulators are demanding that generative AI outputs comply with applicable laws and regulations. But LLMs are computers—they don’t have ethics, and they don’t automatically produce content that adheres to commonly acknowledged ethical standards or applicable laws. Consumers need guardrails that can help balance AI innovation with the need for remaining compliant and mitigating risks. Saifr can help fulfill this role.
Saifr can help address the safety layer for various use cases. First, Saifr can find potential regulatory and reputational risks in financial services marketing documents, whether they were created by AI, humans, or both in collaboration. Saifr flags possible risks, explains why they were flagged, suggests more compliant language, and serves up potentially applicable disclosures. We can also act as a safety layer that scans electronic communications, in real time or in archives, to find instances of quid pro quo, bribery, insider trading, and many other indications of behavior that might be non-compliant with applicable laws in regulated industries. Saifr also uses AI to help uncover potential financial crimes and other crimes that are publicly known via our adverse media screening for AML/KYC solution. In these three ways, Saifr can act as an additional guardrail between consumers, AI, and applicable regulations.
With this announcement, we are making some of our AI models available in the Microsoft Azure AI model catalog for others in the financial services industry. Using our models, a traditional financial services company, a FinTech company, or an app developer using agentic AI to create apps for financial services can use our models to help keep consumers safe. Saifr’s models can now provide an important safety guardrail to even more of the industry. While it will always be important to keep a human in the loop to make final decisions about compliance, we believe this collaboration can lead to wider AI adoption because it helps ease compliance with regulations intended to protect consumers.
There’s a saying: With great power comes great responsibility. AI and language models fall into this category—they can be exceptionally powerful, beneficial tools—but they must be used responsibly and safely. Our mission at Saifr is to help firms take advantage of AI’s amazing capabilities in a safe, ethical way through human augmented compliance.
The opinions provided are those of the author and not necessarily those of Fidelity Investments or its affiliates. Fidelity does not assume any duty to update any of the information. Fidelity and any other third parties are independent entities and not affiliated. Mentioning them does not suggest a recommendation or endorsement by Fidelity.
1178094.4.0
