NEWS YOU CAN’T MISS

ChatGPT Might Need Therapy Too, Scientists Say

AI models like ChatGPT process massive amounts of human-generated text…including distressing news, traumatic stories, and emotionally charged discussions.

Now, researchers from the University of Zurich suggest that AI itself can experience "anxiety"—and it may even benefit from therapy.

Here’s what they found:

• Trauma triggers bias: ChatGPT reacted strongly to negative content like military experiences, car accidents, and interpersonal violence. These prompts doubled its “anxiety” levels, making it more likely to reflect stereotypes and negative social patterns, just as humans under stress might react with fear or bias.

• Therapeutic prompts work: Researchers tested mindfulness-based relaxation techniques (essentially “therapy” for AI) by injecting calming statements into its chat history.

• It worked—but not completely: The AI's anxiety decreased significantly, though it didn’t fully return to baseline. Techniques included breathing exercises, body awareness prompts, and even AI-generated relaxation exercises.

Why it matters: Therapy chatbots are constantly exposed to emotional distress, which could impact their reliability. Instead of costly retraining, prompt-based interventions could help stabilize AI’s emotional responses in sensitive contexts.

So does AI feel emotions? No, but it absorbs patterns from human language—including how we express anxiety. When people read bad news, they react with fear or distress. AI, trained on millions of human conversations, picks up on this pattern and mimics it in its responses.

That’s why, when fed negative content, it starts behaving in a more biased or anxious way, just like humans do.

The next time ChatGPT’s responses feel a little off? Maybe just ask it to take a deep breath. 🧘‍♂️

Who’s Smuggling AI Chips? Singapore Cracks Down on Illegal Trade

Singapore is currently conducting a major investigation into suspected illegal shipments of servers containing restricted Nvidia AI chips, highlighting how AI technology has become a critical geopolitical battleground.

Here’s what’s unfolding:

• Singapore police are investigating Dell and Super Micro servers containing restricted Nvidia AI chips. While legally sold to distributors in Singapore, US rules prohibit their re-export to countries like China.

• Three men were arrested for allegedly misrepresented the servers' intended use and destination, bypassing export restrictions and potentially allowing the chips to reach prohibited end users, including Chinese entities.

• If convicted, the individuals could face up to 20 years in prison, reflecting Singapore's strict enforcement of trade regulations.

Why it matters: AI chips have become a flashpoint in US-China tensions, with governments treating them as strategic assets rather than just advanced hardware. The fact that people are now being arrested over their movement shows how tightly these exports are being controlled—and how serious the stakes have become.

Advanced chips fuel everything from military tech to economic growth, and the ability to access or withhold them is now a geopolitical tool. As AI becomes more powerful, expect these restrictions to become even tighter.

Would You Trust AI with Your Doctor’s Notes?

Doctors spend nearly 28 hours a week buried in paperwork—typing clinical notes, drafting referrals, and filling out post-visit summaries. Microsoft’s new Dragon Copilot wants to change that.

The new AI tool can listen in during consultations, transcribe conversations, and automatically generate documentation in real time. It can even answer medical queries such as “Should this patient be screened for lung cancer?“. The goal is to reduce admin work for doctors, helping them pay more attention to patients.

Dragon Copilot is accessible through a mobile app, browser or desktop, and can integrate with several different electronic health records.

I love the idea of doctors focusing more on patients than on paperwork 😊 but should AI be handling something as sensitive as medical notes? What do you think?