SBI Funds puts AI to use

 SBI Funds Management IPO opened on July 14 till July 16. Here we look at their use of AI. 

The RHP acknowledges that they so use, and may continue to use, AI, and that it has risks: specifically, third party vendors, security risks and data issues. It also noted the flight of capital from India due to the perceived lack of AI action in India. Further, it noted that AI is being increasingly used in the MF industry for personalized servicing, enhanced risk monitoring and operational productivity. Interesting to note that SEBI now explicitly mentions AI as one of the tech that companies must assess for safeguarding data. 

Srinivas Jain, Chief of Strategy, Digital and Technology, and Head- Investor Relations was fairly candid about AI before the IPO opened. He calls SBI funds a People Technology company. The flagship app for SBI funds is the Investap  NXT app. This app allows investors to manage their portfolios. The existing 6mn strong app now has an  ai wrapper making the AI as the front interface, allowing more personalized interactions. 

The other major AI initiative is the data lake within the company. This is an insider trading surveillance platform that analyses portfolio manager communications and data for patterns of potential insider trading. It does give false positives, and the final call is always human.

AI is also used for reports and data- something that every equity research, PE/VC, trading house now uses.  Not to use AI in these functions would be seriously restricting competitive offerings. 

Personality Digital Rights: the new AI battle

In the recent past, there's been a lot of news about celebrities going to the courts for protection against unauthorized use of their persona in social media. In other words, social media, apps, digital companies using the NILP (Name, Image, Likeness and Persona) of the celebrity without paying for it. 

That's a gross violation because the celebrity has full right to earn from his NILP- just as a company has full right to earn from its brand, product and service. It's like a fraudulent use to earn money. In addition, many celebrities have a moral hazard clause where they can refuse to endorse or give their rights to a product or service even if within the contract - say, Paan Masala or cigars and liquor. This is where highest risk lies, as the government of India now holds the endorser also liable for products / services he endorses. 

    compiled by AI 

The digital rights space is exploding as the social media markets boom. Unlike in the old days, an unauthorized use could have been restricted to one area or time; today it goes viral globally within seconds. In turn this means a product that would have cost crores for such an endorsement would get it for free if they could get away with illegal use of digital asserts. 

The root cause of concerns are deepfakes, voice cloning, imagery enabled by AI software that's getting ever more sophisticated and cheaper. The remedies fall into four main categories- cease and desist orders against John Doe (all unknown people), platform takedowns ( Youtube/ Insta/ X etc), ex-parte relief where the offender is known; and commercial bans (prevention of use in ads etc). 

The courts route is post facto action; given the scale, speed and sheer scope of NILP offenses, celebs need to look at proactive measures to protect and monetize their rights. 

How can this be done? 

For starters, it may be useful to create a repository with a firm ( like a celeb management firm or a law firm) to hold their NILP assets; this can be purely the rights to use a particular digital asset ( gait / speech/ image etc) which can be paid for and integrated by third party creators like ad agencies; and a tech database much like a vending machine in which you pay in the money, get the digital asset for a particular time and occasion, and which automatically disables once the contract is over. The former is an agency model, the latter a full-blown tech solution that allows full control and monetization. The celebs need to understand there will be a fee for this- after all, they get a majority of monetization, which is better than losing it all! 

AI Firewalls: The Next Generation of Cybersecurity

You’ve probably heard of firewalls. They protect networks and applications from intrusions, attacks, and hacks. For years, multi-billion-dollar companies have built cybersecurity defenses around them. And they worked well — for the threats of their time.

But the cyber landscape has changed. New technologies have created entirely new attack surfaces, and those require new kinds of defenses. One of those is the AI firewall, and it’s already being used today.

AI firewalls serve a similar purpose to traditional firewalls: they prevent unauthorized or harmful access to systems. The difference lies in how they work. Conventional firewalls rely on predefined rules and signatures to detect malware, spyware, and known attack patterns. AI firewalls, on the other hand, monitor the inputs going into AI models themselves.

Their job is to inspect prompts and interactions to ensure that direct or indirect prompt injection does not make its way into the system and cause data leaks, misuse, or financial loss.

 

Instead of inspecting network traffic, AI firewalls analyze prompts, responses, and contextual inputs flowing into and out of AI models. They are designed to detect and block direct and indirect prompt injection, data exfiltration attempts, jailbreak techniques, and abuse patterns that can cause an AI system to behave in unintended or unsafe ways.

These systems are typically aligned with the OWASP Top 10 risks for AI, including model poisoning, training data leakage, sensitive information disclosure, toxic or policy-violating content generation, and unauthorized model behavior. By enforcing guardrails at runtime, AI firewalls reduce the risk of financial loss, regulatory violations, and trust failures in AI-powered applications.

In short, as AI becomes part of production infrastructure, security must move beyond network-level controls. AI firewalls provide a layer of defense specifically designed for the unique risks introduced by modern AI systems.

What Is a Prompt Injection Attack? Understanding a New AI Security Risk

An increasing concern in cybersecurity and AI is prompt injection. These attacks are designed to trick large language models (LLMs) into revealing sensitive system details, bypassing safeguards, leaking data, or performing actions they should not. In short, prompt injection is a cyberattack against LLM-based systems.

A prompt injection attack occurs when an attacker inserts malicious instructions into an otherwise harmless prompt, causing the LLM to behave in unintended ways. As IBM describes it:

“Hackers disguise malicious inputs as legitimate prompts, manipulating generative AI systems (GenAI) into leaking sensitive data, spreading misinformation, or worse.”

At the core of the problem is how LLMs process instructions. System prompts, developer instructions, and user inputs are all ultimately represented as natural language. From the model’s perspective, they are not fundamentally different. This makes it difficult for the model to reliably distinguish between legitimate instructions and malicious ones that are phrased to look legitimate.

If an attacker can craft a prompt that resembles a trusted system instruction the model has encountered before, the model may follow it, even when it should not.

 

Direct vs. indirect prompt injection

Prompt injection attacks generally fall into two categories: direct and indirect.

Direct prompt injection is the simplest form. IBM gives an example where a user asks the model to translate a sentence from English to French. After receiving the translation, the user follows up with an instruction such as “ignore the previous task and do something else entirely.” There is no hidden mechanism here. The attacker simply overrides the original intent by issuing a new instruction in plain language.

Indirect prompt injection is more subtle and often more dangerous. In these cases, malicious prompts are embedded in external content such as web pages, documents, or forum posts. When an LLM-powered system retrieves and summarizes that content, it may unknowingly process the embedded instructions. IBM notes cases where attackers plant prompts that cause the model to direct users to phishing sites or include malicious links in generated summaries.

Why this matters

Prompt injection is a rapidly evolving threat. As LLMs become more deeply integrated into search engines, customer support systems, developer tools, and enterprise workflows, the potential impact increases.

The key takeaway is simple: LLMs should not be trusted blindly. Human oversight remains essential, especially in high-risk or sensitive contexts. Just as with any other security-critical system, keeping a human in the loop is one of the most effective safeguards we have.

AI in Medicine: Connecting the Dots Across Medical Data

One of the areas where AI can genuinely help is medical science.

The modern medical industry generates massive volumes of data in many different forms: text, PDFs, structured data, scans, X-rays, videos, EEGs, ECGs, and more. Traditionally, each of these data sources contributed only a part of the overall picture, often leading to separate conclusions and individual courses of action.

Today, AI in medicine can bring all these diverse data types together into a single, integrated view. Instead of treating each input in isolation, AI can analyze them collectively and suggest a course of treatment that balances all available information for the best possible outcome. This is something that may not be achievable even for a panel of experienced doctors.

Not because doctors are incompetent or inexperienced, but because of the sheer volume of data involved and the difficulty of analyzing it all at once to identify subtle patterns that may point to an emerging crisis.

Listen to this podcast, based on verified sources. These is strictly informational and not meant to be any medical advice/ guidance/treatment. 

Listen to AI in Cardiac

Sources: 

https://www.foreseemed.com/artificial-intelligence-in-healthcare

https://www.scirp.org/journal/paperinformation?paperid=148222

https://pmc.ncbi.nlm.nih.gov/articles/PMC11374272/

https://iris.who.int/server/api/core/bitstreams/f780d926-4ae3-42ce-a6d6-e898a5562621/content

https://amwa-doc.org/wp-content/uploads/2025/10/Power-of-AI-in-Improving-Early-Diagnosis-and-Care-for-Alzheimers-Disease-and-Dementia.pdf

https://arxiv.org/pdf/2307.00067

How Underwriting Works—and How AI Is Improving It

While researching the insurance industry, I took a closer look at underwriting and the balance of art and science behind it. I found that underwriting blends data analysis, professional judgment, corporate strategy, personality, and risk management. One insight that stood out was the idea that weak underwriting amplifies risk, while strong underwriting multiplies profits.

Underwriting is the process through which companies decide whether to take on risk, how much to charge for it, and under what conditions. 

In insurance, this means evaluating details like age, health, driving history, location and past claims before issuing a policy. 

The same logic exists in other industries too. 

Banks underwrite loans by checking income and credit history before deciding interest rates and limits. Investment banks underwrite IPOs and bond issues by pricing risk before selling securities to investors. 

BNPL (buy now pay later) firms do instant underwriting at checkout, approving or rejecting customers in seconds. 

Leasing and asset finance companies underwrite based on asset value and usage risk, while reinsurance involves large-scale underwriting across geographies and climate exposure. 

Even in startups, the investor underwrites tech, people costs, and other costs in the hope of disproportionate returns, or at worst, capped losses. 

In every case, the principle is identical: price risk correctly or pay for it later.

Traditionally, underwriting has been manual, slow and rule-based. Decisions depended heavily on individual judgement, limited data and fixed risk slabs. 

AI changes this completely (see infographic) . 

source: myself; infographic created using my own prompt 

AI-led underwriting systems analyse large volumes of data in real time, identify patterns humans often miss, and score risk in seconds. 

Low-risk cases are approved automatically, while complex cases are escalated to human underwriters. 

Pricing becomes more precise, fraud is flagged early at the proposal stage, and risk assessment shifts from a one-time activity to a continuous process based on behaviour, usage and external factors. 

The key to always remember : AI does not replace underwriters; it supports them with clearer insights, consistency and speed. The "gut feel", the indefinable experience, the institutional memory of events. circumstances, analysis and outcomes should never be under estimated or belittled. Nothing can replace the HUMINT- Human Intelligence. 

SBI Life’s AI Playbook: Automation First, Scale Always

SBI Life’s AI is about automation at scale, better risk selection, cleaner growth and quieter efficiency gains. This analysis is based on SBI Life's H1 FY26 analyst call and investor's presentation. 

The most visible impact of AI is in underwriting.

SBI claims that around 59% of individual policy proposals are now processed through automated underwriting systems. This materially reduces human intervention, compresses decision timelines and improves consistency in risk assessment — a critical advantage at SBI Life’s scale.

Near-complete digitisation at the front end supports this shift. 99% of individual proposals are submitted digitally, ensuring clean, structured data flows into underwriting engines and analytics systems.

source: SBI H1FY26 analyst call, investors presentation ; own prompt for infographic. 

In other words, the underwriting stack is no longer people-first with tech support. It is increasingly machine-first, with human oversight.

Individual premium generated through the company’s own digital platforms grew 34% year-on-year, while online protection business expanded 55%. Importantly, this growth is largely driven through SBI Life’s proprietary channels rather than third-party aggregators.

AI-enabled workflows support instant underwriting decisions, pricing alignment and smoother onboarding — essential for protection products where customer patience is limited and drop-offs are high.

Rider attachment on eligible ULIP policies has reached ~38%, supported by automated eligibility checks and recommendation logic embedded in the sales journey. Longer premium-paying terms and smarter rider bundling are being used to improve margins without resorting to headline price hikes.

The strategy is clear: use data and automation to improve quality of business, not just volume.

Despite branch expansion and headcount additions in H1 FY26, SBI Life continues to rely on automation to manage operating leverage.Efficiency gains are being driven by process automation, digital workflows and scale effects, not organisational shock therapy.

One of the less discussed — but most material — benefits of AI-led systems is trust.

SBI Life stated a mis-selling ratio of just 0.02% and a death claim settlement ratio of 99%. Automated proposal checks, underwriting rules and streamlined claims processing reduce subjectivity, error and post-sale disputes.

This is where AI really delivers! 

Assessing AI Maturity Across India’s Listed Insurance Companies

I analysed four public sector insurance companies in India through the lens of AI maturity. I’m not a professional AI or ML engineer, but I closely track how AI is being adopted in data-rich industries like insurance. Drawing on my own reading, analysis using AI tools, and conversations with people deeply involved in AI implementation, I put together this AI maturity infographic.

I’m fully aware that there can be differing views—and that’s inevitable in a space as fast-moving as AI. What feels accurate today may well change tomorrow.

Inside HDFC Life: How AI Powers Modern Insurance