AI Explained

AI in Investing: Should You Let AI Pick Your Stocks? What 2026's Data Actually Says

Most investors now ask ChatGPT or Gemini what to buy. A new study shows what happens next – plus Robinhood's AI trading agents and SEBI's coming AI rules.

Key takeaways

  • Using AI for investing is now mainstream: surveys in 2026 find that somewhere between half and three-quarters of retail investors have used AI tools to help them invest.
  • A new study of about 1,900 investors found the opposite of what you might expect: after they started asking an AI assistant, they bought fewer single stocks and more index funds, and their portfolios became better diversified.
  • AI can now trade for you too – Robinhood lets AI agents trade from a separate account – while SEBI says it will soon issue rules for AI in Indian markets, with humans kept in the loop.

Not investment advice. TechPrana explains technology. Nothing here is a recommendation to buy or sell anything. For decisions about your money, talk to a SEBI-registered investment adviser.

Search YouTube for “AI trading” this month and you’ll find videos with hundreds of thousands of views promising that ChatGPT or Claude can pick winning stocks, or that an AI bot “made me money”.

So what actually happens when ordinary investors start asking AI what to buy? In November 2026, the first big study using real portfolios gave a surprising answer.

How many people use AI to invest?

A lot – and it happened fast.

  • An Investing.com survey of 938 US retail investors found 62% already use AI tools to help with investment decisions, mostly chatbots like ChatGPT.
  • Other 2026 surveys put the number even higher: 54% of UK investors (Opinium), 70% across 18,000 people in 26 countries (Amundi) and 73% (HSBC and Ipsos).

In other words, for most new investors, “asking AI” is now as normal as googling a company.

The surprise: AI made investors more boring

Researcher Valerie Baldinger studied about 1,900 investors on Mezzi, a US wealth-management app whose built-in AI assistant answers questions using ChatGPT and Gemini models. She had their real brokerage and retirement account transactions, plus nearly 70,000 questions they asked the AI – before and after they started using it.

The AI couldn’t trade for them. It could only answer questions. Here’s what changed:

  • What people asked for vs. what AI suggested. Investors asked about buying funds in just 3.5% of their questions. But when a question didn’t name a particular stock, 89.7% of the AI’s buy suggestions were ETFs or mutual funds.
  • They followed the advice. After they started using the AI, the share of their monthly purchases going into passive funds rose by about 7.6 percentage points, and the share going into single stocks fell by about 5.9 points.
  • Their portfolios got better diversified. The investors who were least diversified before gained the most – the study estimates they cut the return they were giving up by under-diversifying by about 1 percentage point a year (101 basis points).
  • Slightly cheaper funds. The funds they bought had slightly lower fees.
  • No trading frenzy. There was no significant change in how often they traded.

So instead of turning people into day traders, the chatbots mostly nudged them towards the “boring” diversified investing that most textbooks recommend.

The catches

Before you hand your portfolio to a chatbot, the study comes with important caveats:

  1. These were wealthier investors – the median portfolio was about $556,000 – so the results may not apply to everyone.
  2. The AI’s behaviour depends on its instructions. When the app changed the AI’s system prompt, its fund recommendations went up and down – and investors’ purchases followed. When advice was discouraged, the shift towards funds faded. Whoever writes the instructions shapes the advice.
  3. Everyone gets similar answers. The 10 most-recommended tickers made up about half of all the AI’s ticker recommendations. If millions of people ask the same few models, they may all pile into the same things.
  4. It’s a model of risk, not actual returns. The study measures diversification, not whether people ended up richer.

Now AI can trade for you, too

Chatbots answering questions is one thing. In 2026, AI started placing trades.

  • On 27 May 2026, Robinhood launched Agentic Trading in beta, letting customers connect their own AI agents (such as ones built on ChatGPT or Claude) to a brokerage account. The agent gets a separate account, and can only use money you deliberately put into it. It started with US stocks only.
  • More than 70,000 AI agent accounts were created in the beta.
  • In September, Robinhood went further with Robinhood Agents, which help customers build strategies, do research or trade automatically.

Even Robinhood warns about the risks: its disclosure says agents can misunderstand instructions, act on stale data and behave unpredictably – and that the customer is responsible for what the agent does.

This isn’t available in India in the same form yet.

What about India? SEBI is writing the rules

India’s market regulator is moving on AI from several directions:

  • AI rules are coming. On 14 November 2026, SEBI chairman Tuhin Kanta Pandey said SEBI will issue detailed guidelines on how AI is used in capital markets. They will cover disclosures, accountability and human oversight, with a “human-in-the-loop” approach: critical decisions made with AI’s help would still need human supervision. No date has been given yet.
  • AI-generated advice is still advice. Under SEBI’s rules, firms can’t blame an algorithm. If a registered adviser uses AI and the advice is wrong, the adviser is responsible. And if a course or channel gives buy/sell calls, it counts as investment advice even if it’s labelled “education”.
  • SEBI uses AI too. The regulator says it now uses AI to track financial influencers giving illegal investment advice on social media, in real time.

That’s worth remembering the next time a video promises an “AI stock-picking system”.

How to use AI for investing – safely

AI can be genuinely useful as a research assistant. Here’s a sensible way to use it:

  • ✅ Use it to understand, not to decide. Ask it to explain an annual report, what an expense ratio is, or how an index fund works.
  • ✅ Check every number. Chatbots can get prices, dates and figures wrong. Verify on the exchange (NSE/BSE) or the company’s own filings.
  • ✅ Ask for the risks. “What could go wrong with this?” is often a better question than “Should I buy this?”
  • ✅ Remember it doesn’t know you. AI doesn’t know your income, goals, loans or risk appetite unless you tell it – and even then it isn’t a registered adviser.
  • ❌ Never share passwords, OTPs or account numbers with a chatbot.
  • ❌ Don’t pay for “AI stock tips” from unregistered sellers. Check the SEBI registration of anyone giving you buy/sell calls.
  • ❌ Don’t let an AI agent trade with money you can’t afford to lose.

The bottom line

The early evidence is more reassuring than the hype: when ordinary investors ask AI for help, the best-studied chatbots tend to push them towards diversification and low-cost funds, not wild bets. But the advice depends on how the AI is set up, everyone is getting similar answers, and AI agents that trade on their own are brand new.

Use AI like a smart, fast research assistant – and keep the final decision (and the responsibility) with you, ideally with a SEBI-registered adviser.

Want to work in this space? AI, fintech and data skills are in demand – check your skills on Career Genie.

Sources

  1. Valerie Baldinger: Generative AI and Retail Investment (working paper, November 2026)
  2. ETF Stream: AI chatbots steer retail investors towards index funds, study finds (19 Nov 2026)
  3. Investing.com: How retail investors are using AI in 2026 (survey of 938 US investors)
  4. SiliconANGLE: Robinhood opens its platform to AI agents for trading (27 May 2026)
  5. Yahoo Finance: Robinhood opens platform to AI agents
  6. PYMNTS: Robinhood unveils Robinhood Agents at HOOD Summit (29 Sep 2026)
  7. Crypto Briefing: More than 70,000 AI agent accounts created in Robinhood's beta
  8. Angel One: SEBI to issue AI usage guidelines for capital markets (16 Nov 2026)
  9. YourStory: SEBI to issue AI usage guidelines, focus on investor protection
  10. Kotak Neo: SEBI uses AI to track financial influencers in real time (Feb 2026)
  11. Mondaq (Aarna Law): SEBI's 2026 rules – buy/sell calls labelled as 'education' are still advice
AIInvestingStock MarketChatGPTSEBIIndiaRobinhood

Keep reading