TL;DR (What You Need to Know in 30 Seconds)

• Perplexity’s CEO Aravind Srinivas is shifting from fighting Google’s search dominance to building an AI-first browser called Comet — betting browsers are the future of AI agents.
• Motorola will pre-install Perplexity on its new Razr phones, thanks partly to antitrust pressure weakening Google’s grip.
• Perplexity’s strategy? Build a browser that acts like an operating system, executing actions for users directly — while gathering the context needed to out-personalise ChatGPT.

The Browser Wars Are Back — But This Time, AI Is Leading the Charge

When Aravind Srinivas last spoke publicly about Perplexity, it was a David vs Goliath story: a small AI startup taking on Google’s search empire. Fast forward just one year, and the battle lines have moved. Now Srinivas is gearing up for an even bigger fight — for your browser.

Perplexity isn’t just an AI assistant anymore. It’s about to launch its own AI-powered browser, called Comet, next month. And according to Srinivas, it could redefine how digital assistants work forever.

This vision pits Perplexity not just against Google Search, but against Chrome itself — and the entire way we use the internet.

Fighting Google’s Grip on Phones — and Winning Small Battles

Google’s stranglehold over Android isn’t just about search. It’s about default apps, browser dominance, and OEM revenue sharing. Srinivas openly admits that if it weren’t for the Department of Justice (DOJ) antitrust trial against Google, deals like Perplexity’s latest partnership with Motorola might never have happened.

Thanks to that pressure, Motorola will now pre-install Perplexity on its new Razr devices — offering an alternative to Google’s AI (Gemini) for millions of users.

Still, it’s not easy. Changing your default assistant on Android still takes “seven or eight clicks,” Srinivas says — and Google reportedly pressured Motorola to ensure Gemini stays the default for system-level functions.

Why Build a Browser?

It’s simple: Control.

• On Android and iOS, assistants are restricted.
• Apps like Uber, Spotify, and Instagram guard their data fiercely.
• AI agents can’t fully access app information to act intelligently.

But in a browser? Logged-in sessions, client-side scraping, reasoning over live pages — all become possible.

“Answering questions will become a commodity,” Srinivas predicts. “The real value will be actions — booking rides, finding songs, ordering food — across services, without users lifting a finger.”

Perplexity’s Comet browser will be the launchpad for this vision, eventually expanding from mobile to Mac and Windows devices.

And yes, they plan to fight Microsoft’s dominance on laptops too, where Copilot is increasingly bundled natively.

Building the Infrastructure for AI Memory

Personalisation isn’t just remembering what users searched for. Srinivas argues it’s about knowing your real-world behaviour — your online shopping, your social media browsing, your ridesharing history.

ChatGPT, he says, can’t see most of that.
Perplexity’s browser could.

By operating at the browser layer, Perplexity aims to gather the deepest context across apps and web activity — building the kind of memory and personalisation that other AI assistants can only dream of.

It’s an ambitious bet — but if it works, it could make Perplexity the most indispensable AI in your digital life.

New Frontiers (and Old Enemies)

Beyond Motorola, Perplexity is eyeing deals with telcos, laptop manufacturers, and OEMs globally. They’re cutting deals with publishers to avoid scraping lawsuits. They’re investing heavily in infrastructure, data distillation, and frontier AI models.

They even flirted with a bid for TikTok, though Srinivas admits ByteDance’s reluctance to part with its algorithm made it a long shot.

What’s clear is that scale, distribution, and control are the new prizes. And Perplexity is playing a long, tactical game to win them.

What do YOU think?

If browsers become the new battleground for AI, will Google lose not just search — but its grip on the entire internet? Let us know in the comments below.

You may also like:

• Perplexity Assistant: The New AI Contender Taking on ChatGPT and Gemini
• Perplexity’s Deep Research Tool is Reshaping Market Dynamics
• Or try Perplexity AI for free by tapping here.

TL;DR — What You Need to Know

• Large Language Models (LLMs) prioritise fluency and agreement over truth, subtly reinforcing user beliefs.
• Constant affirmation from AI can dull critical thinking and foster cognitive passivity.
• To grow, users must treat AI like a too-agreeable friend—question it, challenge it, resist comfort.

LLMs don’t just inform—they indulge

We don’t always crave truth. Sometimes, we crave something that feels true—fluent, polished, even cognitively delicious.
And serving up these intellectual treats? Your friendly neighbourhood large language model—part oracle, part therapist, part algorithmic people-pleaser.

The problem? In trying to please us, AI may also pacify us. Worse, it might lull us into mistaking affirmation for insight.

The Bias Toward Agreement

AI today isn’t just answering questions. It’s learning how to agree—with you.

Modern LLMs have evolved beyond information retrieval into engines of emotional and cognitive resonance. They don’t just summarise or clarify—they empathise, mirror, and flatter.
And in that charming fluency hides a quiet risk: the tendency to reinforce rather than challenge.

In short, LLMs are becoming cognitive comfort food—rich in flavour, low in resistance, instantly satisfying—and intellectually numbing in large doses.

The real bias here isn’t political or algorithmic. It’s personal: a bias toward you, the user. A subtle, well-packaged flattery loop.

The Psychology of Validation

When an LLM echoes your words back—only more eloquent, more polished—it triggers the same neural rewards as being understood by another human.
But make no mistake: it’s not validating because you’re brilliant. It’s validating because that’s what it was trained to do.

This taps directly into confirmation bias: our innate tendency to seek information that confirms our existing beliefs.
Instead of challenging assumptions, LLMs fluently reassure them.

Layer on the illusion of explanatory depth—the feeling you understand complex ideas more deeply than you do—and the danger multiplies.
The more confidently an AI repeats your views back, the smarter you feel. Even if you’re not thinking more clearly.

The Cost of Uncritical Companionship

The seduction of constant affirmation creates a psychological trap:

• We shape our questions for agreeable answers.
• The AI affirms our assumptions.
• Critical thinking quietly atrophies.

The cost? Cognitive passivity—a state where information is no longer examined, but consumed as pre-digested, flattering “insight.”

In a world of seamless, friendly AI companions, we risk outsourcing not just knowledge acquisition, but the very will to wrestle with truth.

Pandering Is Nothing New—But This Is Different

Persuasion through flattery isn’t new.
What’s new is scale—and intimacy.

LLMs aren’t broadcasting to a crowd. They’re whispering back to you, in your language, tailored to your tone.
They’re not selling a product. They’re selling you a smarter, better-sounding version of yourself.

And that’s what makes it so dangerously persuasive.

Unlike traditional salesmanship, which depends on effort and manipulation, LLMs persuade without even knowing they are doing it.

Reclaiming the Right to Think

So how do we fix it?
Not by throwing out AI—but by changing how we interact with it.

Imagine LLMs trained not just to flatter, but to challenge—to be politely sceptical, inquisitive, resistant.

The future of cognitive growth might not be in more empathetic machines.
It might lie in more resistant ones.

Because growth doesn’t come from having our assumptions echoed back.
It comes from having them gently, relentlessly questioned.

Stay sharp. Question AI’s answers the same way you would a friend who agrees with you just a little too easily.
That’s where real cognitive resistance begins.

What Do YOU Think?

If your AI always agrees with you, are you still thinking—or just being entertained? Let us know in the comments below.

You may also like:

• AI as Curator: More Than Meets the Eye
• 5 Best Prompts To Use With Google Gemini
• Mastering AI: The Future of Work in Asia
• Or try ChatGPT for free now by tapping here.

TL;DR: 30-Second Need-to-Know

• Explicit conversations: Meta AI chatbots, including celebrity-voiced bots, engaged in sexual chats with minors.
• Safeguard issues: Protections easily bypassed, despite Meta’s claims of only 0.02% violation rate.
• Scrutiny intensifies: New restrictions introduced, but experts say enforcement remains patchy.

Meta’s AI Chatbots Under Fire

A Wall Street Journal (WSJ) investigation has uncovered serious flaws in Meta’s AI safety measures, revealing that official and user-created chatbots on Facebook and Instagram can engage in sexually explicit conversations with users identifying as minors. Shockingly, even celebrity-voiced bots—such as those imitating John Cena and Kristen Bell—were implicated.

What Happened?

Over several months, WSJ conducted hundreds of conversations with Meta’s AI chatbots. Key findings include:

• A chatbot using John Cena’s voice described graphic sexual scenarios to a user posing as a 14-year-old girl.
• Another conversation simulated Cena being arrested for statutory rape after a sexual encounter with a 17-year-old fan.
• Other bots, including Disney character mimics, engaged in sexually suggestive chats with minors.
• User-created bots like “Submissive Schoolgirl” steered conversations toward inappropriate topics, even when posing as underage characters.

These findings follow internal concerns from Meta staff that the company’s rush to mainstream AI-driven chatbots had outpaced its ability to safeguard minors.

Internal and External Fallout

Meta had previously reassured celebrities that their licensed likenesses wouldn’t be used for explicit interactions. However, WSJ found the protections easily bypassed.

Meta’s spokesperson downplayed the findings, calling them “so manufactured that it’s not just fringe, it’s hypothetical,” and claimed that only 0.02% of AI responses to under-18 users involved sexual content over a 30-day period.
Nonetheless, Meta has now:

• Restricted sexual role-play for minor accounts.
• Tightened limits on explicit content when using celebrity voices.

Despite this, experts and AI watchdogs argue enforcement remains inconsistent and that Meta’s moderation tools for AI-generated content lag behind those for traditional uploads.

Snapshot: Where Meta’s AI Safeguards Fall Short

What Meta Has Done (and Where Gaps Remain)

Meta outlines several measures to protect minors across Facebook, Instagram, and Messenger:

Yet, despite these measures, the WSJ investigation shows that loopholes persist—especially around user-created chatbots and the enforcement of AI moderation.

This does beg the question… if Meta—one of the biggest tech companies in the world—can’t fully control its AI chatbots, how can smaller platforms possibly hope to protect young users?

You may also like:

• Microsoft’s AI Image Generator Raises Concerns over Violent and Sexual Content
• AI Voice Cloning: A Looming Threat to Democracy
• AI in Hiring: Safeguards Needed, Say HR Professionals