Thursday, October 8, 2026

TikTok ran experiments in which thousands of users — including children and teenagers — were allegedly given non-working versions of safety features, according to allegations in a newly unsealed New York lawsuit.

One example was Algo Refresh, a feature designed to reset a user's recommendations when they want to get away from unwanted or potentially harmful content.

According to the lawsuit, some users believed they had switched the feature on — but their feeds didn't actually change.

The allegations say TikTok used these “ghost” or placebo versions to study the effect on user engagement and advertising revenue. TikTok denies the allegations and says protecting young users is a priority.

A safety feature only matters if it actually works.

The bigger question here isn't just what happened inside TikTok's experiment. It's whether users can trust a platform when they believe they've taken a step to protect themselves — especially when the users are children.

Samsung expects its third-quarter operating profit to hit 107.4 trillion won — about $80 billion.

That's a staggering 782.5% increase from the same period last year and the first time Samsung has crossed the 100 trillion won mark in quarterly operating profit.

The main engine? AI chips.

Demand for memory chips used in AI infrastructure has sent prices and sales soaring. Samsung's semiconductor business is benefiting from companies around the world spending heavily to build AI data centres.

And there's an important catch: Samsung's consumer electronics and mobile businesses aren't performing nearly as well, showing that the AI boom isn't lifting every part of the company equally.

AI isn't just making software companies richer. It's creating a massive hardware economy underneath them.

Every chatbot, AI agent and data centre needs chips — and Samsung is showing just how profitable that infrastructure race can become.

Cybersecurity company CrowdStrike says a suspected attacker used AI-powered tools during cyberattacks against South Korean financial institutions.

The attacks targeted at least nine financial institutions, with data breaches reported at several banks. CrowdStrike says the attacker used tools including ARTEX, an AI-powered penetration-testing tool, alongside large language models and Claude Code.

South Korean authorities are investigating the attacks. Importantly, CrowdStrike says the identity of the suspected attacker has not been confirmed.

AI doesn't care whether the person using it is building something or breaking into something.

The same technology that can help someone write code faster can potentially help an attacker find vulnerabilities faster.

That means cybersecurity is becoming an AI race too — and the people defending systems may have to move just as quickly as the people attacking them.

AI is becoming infrastructure.

Today's three stories show three very different consequences:

📱 Platforms are being challenged over how they protect users.
💰 Chipmakers are making extraordinary amounts of money from AI demand.
🔐 Attackers are finding new ways to use AI against financial systems.

The AI story is getting bigger than chatbots.

Ask yourself:

Would you know if an app's safety feature wasn't actually working?

That's the uncomfortable question sitting underneath today's biggest story.

Technology is becoming more powerful — but trust is becoming just as important as capability.

AI is making billions, changing platforms and powering new threats — but the real question is whether we can trust the technology we're increasingly relying on.