FAQs
You have questions, we have answers.
AI for compliance
What is RegTech?
Short for regulatory technology, RegTech is the application of technology—including AI, automation, and data analytics—to simplify, enhance, and assist regulatory compliance. RegTech can offer organizations cost-effective, practical ways of improving compliance processes such as fraud detection, stress testing, risk assessment, and much more. Learn more.
Why should I use AI for compliance?
AI can be an extra set of eyes, helping expand and deepen compliance oversight. Its ability to rapidly process huge volumes of data can give compliance teams easy access to real-time analytics and risk reporting for a faster, more streamlined compliance experience. Together, these capabilities can help scale compliance processes beyond human limitations.
What are the different types of AI?
Artificial intelligence (AI) is an umbrella term for many techniques that enable machines to mimic human behavior. Types of AI include:
- Rules-based: The simplest type of AI, consisting of systems where humans define the patterns via rules pre-written by experts.
- Machine learning: Computers learn to recognize complex patterns in data without humans having to explicitly describe all the patterns of interest.
- Natural language processing: A subset of machine learning in which computers understand human language as it’s written and spoken.
- Generative AI: Instead of analyzing something that already exists, computers create new content based on patterns and data they have learned from.
- Large language models: Computers trained on billions of datapoints use existing text (input) to generate new text (output).
- Agentic AI: Systems designed to act as agents and interact smartly and autonomously, invoking other technologies and making decisions.
What are some risks of using AI?
AI risks range in significance, including inaccuracy, model drift, data privacy risks, and lack of model transparency. Most risks associated with AI can be mitigated with appropriate human oversight and appropriate controls.
How were Saifr’s AI models trained?
- SaifrReview and SaifrScan were trained on high-quality, compliance-curated data that represents over 20 years of work by thousands of marketing and compliance experts in various lines of business within the financial services industry.
- SaifrScreen’s text models were trained on publicly available web documents, while its client-specific risk relevancy models are trained on client feedback on the outputs of the text models.
- Saifr eComms’ models were trained on a variety of sources, including open-source and synthetic data.
SaifrRevew’s, SaifrScan’s, and Saifr eComms’ data outputs were validated by subject matter experts—humans in the loop—to help mitigate bias and ensure accuracy.
Which regulations can Saifr help address?
Across three product areas, Saifr helps firms comply with aspects of numerous financial services regulations:
- Public communications: FINRA 2210, SEC 482, SEC Marketing Rule
- Adverse media screening: Anti-money laundering regulations, Know Your Customer expectations, Bank Secrecy Act, U.S. Office of Foreign Assets Control sanctions
- Electronic communications: Bank Secrecy Act; aspects of numerous regulations from U.S. federal law, FINRA, SEC, CFPB, OCC
How does Saifr’s AI stay up to date with regulations?
When regulations change, our compliance and data science teams work together to fine-tune and validate our AI models with our SME team.
How does Saifr use AI ethically/responsibly?
Saifr’s human-centric approach to AI recognizes the need for transparent, explainable, unbiased, and representative systems. There’s a whole team of humans in the loop throughout model development, and the data used to train our SaifrReview, SaifrScan, and Saifr eComms models is thoroughly reviewed by subject matter experts to help ensure it is ethical and accurate. Learn more in our Responsible AI Principles.
How does Saifr protect client privacy and maintain security?
SaifrReview, SaifrScan, and SaifrScreen are SOC 2 Type 2 certified, which involves an annual, independent audit including testing and validating of security and privacy controls.
Can Saifr integrate with existing systems?
Yes, all Saifr’s capabilities are available as APIs.
What industries and regions does Saifr serve?
- SaifrScan, SaifrReview, and Saifr eComms are designed for financial institutions in the U.S.
- SaifrScreen helps U.S. and multinational firms, including those in financial services and other industries, with their AML/KYC and Trust & Safety programs.
How can I get started with Saifr?
Request a demo. We’ll start with a conversation about your needs, then show you how we can help.
Is Saifr owned by Fidelity Investments/Labs?
Saifr is part of Fidelity Labs, Fidelity Investments’ in-house software incubator. Learn more about Labs.
Adverse media screening
What is adverse media screening?
Adverse media screening reviews online and offline sources to uncover information about the data entity that may introduce regulatory and reputational risks to a business. It is important for AML and KYC compliance programs, helping to identify risks related to financial crimes, corruption, fraud, and more. Firms in regulated industries should consider conducting regular checks and monitoring their customer and vendor ecosystems continuously as important aspects of helping to maintain a good reputation and reduce risks related to financial crimes.
What is unstructured vs. structured data?
Unstructured data make up 80% of internet data and are active, real-time data that can provide a wealth of information. Structured data represent just 20% of available data and are found in a predefined format in curated databases that have to be manually updated. Incorporating both unstructured and structured data into AML/KYC processes can dramatically improve the breadth and depth of a firm’s coverage. Learn more.
What is entity resolution?
Entity resolution is the process of predicting that an entity in unstructured data is the same one being screened. It’s a critical part of adverse media screening procedures. Learn about AI-assisted entity resolution.
Why do false positives/negatives matter?
For AML/KYC screening, a false positive is a result that incorrectly indicates the presence of risk. The greater false positives a solution generates, the fewer real risks your team may be surfacing. False negatives are even more dangerous, representing bad actors that have slipped through your defenses. SaifrScreen’s mission is to alert on potential bad actors that may be systematically mistaken for good actors.
What are the AML/KYC/BSA regulations?
- The purpose of anti-money laundering (AML) rules is to help detect and report suspicious activity, such as the offenses that often precede money laundering and terrorist financing: securities fraud, market manipulation, etc.
- Know Your Customer (KYC) guidelines prompt financial institutions and businesses to verify the identity and risks associated with customers (current and potential), aiming to uncover suspicious behaviors like money laundering and financial terrorism before they happen.
- The Bank Secrecy Act (BSA) establishes program, recordkeeping, and reporting requirements for U.S. financial institutions to help government agencies detect and prevent money laundering.
Marketing compliance review
Why is marketing compliance review necessary?
Compliance reviews of marketing content help firms ensure their content is fair, balanced, and not misleading.
What are the FINRA and SEC advertising guidelines?
FINRA Rule 2210 is the primary rule governing broker dealers’ communications with the public. SEC Investment Adviser Marketing Rule 206-4 (aka the Marketing Rule) is the primary rule governing investment advisors’ communications with the public. This ebook explains more.
What communications materials are subject to financial services rules and regulations?
A range of internal or external communications materials, whether they’re for the general public, institutional investors, or a limited number of targeted individuals, can be subject to regulators of the financial industry, such as the SEC or FINRA. Read this ebook to learn more.
Can Saifr scan videos for compliance risks?
Yes, Saifr can transcribe videos and scan the text for potentially risky language, marking where in the video the risk is and explaining how it might not be compliant. Plus, Saifr’s transcription tool is familiar with financial terms and will show “401(k)” and “mutual fund plan”—not “four oh one kay” and “mutual fun planned” like some others might.
How can AI help solve pain points related to complying with advertising regulations?
Natural language processing (NLP)—a type of AI—can read and understand the context of words and phrases, meaning it can recognize which sentences may violate regulatory guidelines and which disclosures might be needed. It can perform a first pass review more quickly and act as an extra set of expert eyes reviewing content, which can result in a cleaner first draft for compliance review, fewer rounds of review, and faster time to market.
Electronic communications surveillance
What is electronic communications surveillance?
Electronic communications (eComms) surveillance is the process of monitoring the communications that occur on business-approved channels (e.g., email, chat, video conferencing) for potential violations of company policy or regulatory risks.
Why is archiving electronic communications important?
The SEC and CFTC require some financial firms to retain and preserve a range of electronic business records (such as certain transaction details, correspondence, and financial documents). These regulations are designed to ensure transparency and accountability to regulators and protection to investors.
What are the electronic communications regulations for financial firms?
There are a wide range of regulations that influence electronic communications for financial firms, including those from the SEC and CFTC governing recordkeeping, plus those from FINRA, the CFPB, the OCC, and more pertaining to the content of the communications.
What are off-channel communications?
Off-channel communications are conversations that happen outside of business-approved platforms. Since some financial firms can be subject to recordkeeping regulations, communications that aren’t appropriately recorded and record kept can pose a compliance risk.
Did we miss one? Drop us a line and we'd be happy to provide an answer.
1175368.1.0