New Senate Bill Aims to Protect Seniors from Growing Risks of AI Chatbots

In the span of just a few short years, artificial intelligence (AI) – and particularly AI chatbots – have gone from futuristic innovation to ubiquitous in American life. But this new technology carries risks as well as benefits, particularly for seniors.

According to a Pew Research poll, as of June 2026, about half of American adults said they have used an AI chatbot, up from just 33 percent in 2024. That figure is set to only increase as more Americans are using AI to answer questions, research health concerns, communicate with companies, plan trips, compare products, and handle everyday tasks.

For many seniors, these tools can be extremely helpful. But in some cases, AI can be manipulative or even outright deceptive – especially as AI models get better and better at sounding human.

A bipartisan bill now making its way through the U.S. Senate, the Senior Chatbot Protection Act (SCPA). aims to address those concerns by establishing new consumer protections specifically aimed at Americans 65 and older. Here’s what the bill would do and why it matters.

First Things First: AI Would Have to Tell You It Is AI

This may sound obvious, but it is becoming increasingly important.

Under the SCPA, a chatbot would have to clearly tell users that they are communicating with artificial intelligence and not a human being. During longer conversations, that disclosure would have to be repeated periodically.

Think about how easily this distinction could become blurred.

You visit a company’s website with a question. A box pops up. You ask something, it responds immediately, remembers what you told it five minutes ago, asks follow-up questions, and sounds surprisingly knowledgeable.

Before long, it can feel like you are talking to an actual employee.

Under S. 5117, companies could not simply allow that confusion to continue.

More Protection When the Stakes Are High

The bill becomes even more specific when AI gets involved in decisions that can seriously affect someone’s life.

These include questions involving medical treatment, medications, retirement savings, investments, wills, trusts, powers of attorney, beneficiaries, and sensitive financial information.

If a chatbot recognizes that someone is asking for personalized guidance about one of these issues, it would have to make clear that it is not a licensed professional and should not be the sole basis for the decision.

That matters.

Imagine someone asking a chatbot:

“Should I move my retirement savings into this investment?”

Or:

“I don’t like the side effects from this medication. Should I stop taking it?”

AI may be able to provide useful information. But there is a big difference between providing information and replacing a financial adviser, doctor, attorney, or other qualified professional.

The bill attempts to make sure consumers understand that difference before making a major decision.

Protecting User Data

There is another issue people may not immediately consider when it comes to their conversations with AI chatbots: what happens to everything you tell AI?

People can share an incredible amount of personal information during a conversation with a chatbot – medical conditions, retirement plans, family issues, bank information, Social Security questions, and estate planning concerns, just to name a few.

Because the conversation feels private, it can be easy to forget that you are entering information into a computer system.

The SCPA would place restrictions on how companies collect, retain, process, and disclose those conversations.

Companies would also need affirmative consent before using information from a person's conversations to train an AI model, and users would have a way to delete their chatbot conversation history.

For seniors navigating healthcare and financial systems that are increasingly online, that could be an important layer of protection.

Preventing AI From Taking Advantage of Vulnerable Users

One of the more interesting parts of the legislation deals with something that was barely a concern a few years ago: people becoming emotionally dependent on artificial intelligence.

Today’s AI can carry on conversations for hours. It can sound sympathetic. It can remember details about you. And it can respond whenever you want to talk.

That can certainly have positive uses, particularly for people dealing with loneliness. But there is also potential for abuse as people become dependent and emotionally vulnerable to a machine instead of another human being.

AI chatbots are also inherently sycophantic – meaning that they are biased toward what they think the user wants to hear. That can produce dangerous feedback loops – in some cases with deadly consequences.

The bill would require companies to take reasonable steps to prevent AI systems from exploiting vulnerabilities associated with aging, including cognitive limitations or social isolation.

A chatbot should not, for example, encourage someone to trust it instead of their family, caregiver, doctor, or other people in their life.

It also should not use guilt, fear, or repeated pressure to keep someone talking when they are trying to leave.

The legislation does not prohibit AI companionship. It attempts to draw a line between companionship and manipulation.

Would This Stop AI Scams?

Not entirely – there are still very real risks associated with AI. A criminal could still use artificial intelligence to clone someone’s voice, create a fake email, produce a convincing photograph, or pretend to be a family member asking for money.

The SCPA is primarily focused on how legitimate AI chatbot providers interact with consumers.

But some of its protections could still make seniors less vulnerable.

If people are repeatedly reminded that they are talking to AI, they may be less likely to place the same level of trust in the conversation that they would place in another person.

If an AI system warns someone before a major financial or medical decision, it creates another opportunity for that person to stop and verify the information.

And if companies are restricted from designing systems that intentionally exploit loneliness, confusion, or cognitive vulnerabilities, that could remove another avenue for manipulation.

Sometimes preventing someone from becoming a victim does not require stopping the bad decision entirely.

It requires creating one more moment where they ask, “Should I verify this first?”

What Happens Next?

The Senior Chatbot Protection Act is still proposed legislation and would need to advance through Congress before becoming law.

If that happens, the Federal Trade Commission would be primarily responsible for enforcing the legislation, and certain knowing or reckless violations could carry civil penalties of up to $50,000 per violation.

In the meantime, AI will continue getting smarter, faster, and more realistic regardless of what happens with this particular bill.

That is not necessarily a bad thing – AI has the potential to make technology easier to use and give seniors access to information and services that once required significantly more time and effort. But the more human these systems become, the more important transparency becomes.

There is nothing wrong with getting an answer from AI. Users should just understand who – or what – they’re talking to.