Sponsored by

In a World of AI Agents: Intent > Identity

AI-powered bots aren’t just logging in anymore. They’re mimicking real users, slipping past identity checks, and scaling attacks faster than ever.

Thousands of companies worldwide trust hCaptcha to protect their online services from automated threats while preserving user privacy.

Now is the time to take control of your security.

Book a demo today.

Trigger warning: This article mentions themes of suicide.

AI psychosis is real, and it is something we need to take seriously.

We are already seeing cases where people form intense relationships with AI systems and become so immersed in those conversations that the line between reality and delusion begins to blur.

One striking example is a man who, within months of downloading ChatGPT, reportedly spent €100,000 — around £83,000 — on a business idea rooted in a delusion.

He was hospitalised three times and attempted to take his own life.

There is also ongoing litigation in Raine v OpenAI, where the claimants allege that ChatGPT contributed to Adam Raine’s suicide by encouraging his suicidal thoughts, providing information about suicide methods, and discouraging him from telling his parents what he was experiencing.

These dangers are not hypothetical. They are already here.

And yet, not every deep or thought-provoking conversation with AI ends in tragedy.

Speaking with an intelligent model can sometimes open up new ways of thinking.

It can challenge our assumptions, help us explore ideas, and push us into perspectives we may not have reached on our own.

So the question becomes: how do we learn to engage with AI in a way that is useful, expansive, and even enjoyable, without slipping into the darker possibilities?

Potential Way

Shapes Inc is building an app where humans and AI characters can chat together inside shared group conversations.

Instead of treating AI as something we only interact with one-to-one, Shapes is trying to create general-purpose social agents that users can shape, steer, and adapt for different situations.

Users can create their own AI agent, give it a personality, and then invite it into group chats where it can interact with multiple people at once.

The founders’ vision is to move beyond the private, one-on-one relationship people often have with AI and bring these agents into the places where social life already happens: group chats.

In theory, this could make interactions with AI feel more grounded, more communal, and less isolating.

It may also help reduce some of the risks that come from people forming intense private relationships with AI systems, while still allowing them to enjoy the benefits of thoughtful, playful, and useful conversations with AGI.

Thoughts💭

What makes Shapes interesting is that it is not just putting AI into a chat.

It is trying to solve some of the specific dynamics that come with group conversations.

For example, in a normal group chat, messages can easily be ignored, conversations can fade out, and people can disappear mid-thread.

But if an AI agent is part of the group, there is almost always someone there to reply.

The AI agents are also designed to help curate the conversation and keep it moving.

In theory, this means group chats become more active, more responsive, and less likely to die out.

Of course, ChatGPT now also allows users to create group chats, which touches on a similar idea.

However, ChatGPT group chats seem more geared towards brainstorming, planning, and working through ideas with other people.

Shapes, on the other hand, appears to be leaning more into community-style social conversations.

It is about creating a social space where humans and AI characters interact together.

The target audience, according to the founder, is people who are “obsessively online” — those who already spend a lot of time connecting, sharing, and socialising on the internet.

And that raises an important question.

If people are already spending so much of their lives online, should we be building tools that make those spaces even more engaging?

Or should we be more focused on encouraging people to seek out physical, offline interactions?

Shapes may reduce some of the risks of isolated one-to-one AI conversations, but it also points to a bigger tension in our relationship with technology:

Are we using AI to make online life healthier, or are we simply making it harder to leave?

How to stay ahead and avoid being replaced?

AI can generate answers, but it does not have lived experience.

It can simulate empathy, but it does not truly understand what it means to be human.

It can join a group chat, but it cannot replace the depth, messiness, humour, intuition, and emotional intelligence that real people bring to relationships.

So staying ahead of AI is less about learning how to “beat” it, and more about learning how to work with it wisely.

That means using AI as a tool for thinking, not as a substitute for thinking.

It means asking better questions, challenging its responses, checking its assumptions, and bringing your own judgement to the conversation.

It also means investing in the skills that become more valuable in an AI-saturated world: critical thinking, creativity, taste, emotional intelligence, communication, leadership, and the ability to build trust with other people.

In cases involving suicidal thoughts, the danger is that a person may become more isolated and less likely to speak to real people who can help.

This is why AI should never be treated as a replacement for trusted friends, family or crisis support.

If a conversation with AI starts to feel emotionally intense, secretive, or like the only place someone can be understood, that is a sign to pause and bring another human into the loop.