TÜRSAB • License No. 4035
Global Highlights: September 17–22, 2026
World & General
The world has been unusually busy lately — although perhaps “unusually” is no longer the right word.
One thing that keeps standing out is how quickly technology is becoming part of modern warfare. Drones were only the beginning. We are now seeing remotely controlled boats, AI-assisted systems and, increasingly, autonomous machines interacting with one another on the battlefield.
One recent example came from the Russia–Ukraine war. A Ukrainian unmanned naval vessel reportedly engaged and sank a Russian remotely operated combat boat. No soldiers were physically inside either vessel.
It sounds like something from a science-fiction film, but it is already part of military reality.
And perhaps the strangest question is this: if machines eventually fight machines, does that make warfare safer for humans — or simply make it easier to wage?
A snapshot from the incident.
A Ukrainian Sargan-3000 drone boat destroyed a Russian naval drone using a remote-controlled 12.7 mm machine gun.
Meanwhile, the economic and technological consequences of conflict continue to spread.
Amazon is still dealing with lasting disruptions at data centers in Bahrain and the UAE following attacks earlier this year. Hundreds of services reportedly remain offline.
The financial side is becoming equally interesting. The U.S. Treasury has sanctioned an Iranian cryptocurrency platform after alleging that Bitcoin was being used as part of a system for collecting payments connected to ships passing through the Strait of Hormuz.
Cryptocurrency was supposed to make money borderless.
Apparently, governments are learning how to make sanctions borderless too.
Speaking of crypto, Sam Altman’s World project has expanded its services this week. The company, which uses its controversial Orb devices to verify people’s identities, has launched World Money, a financial application incorporating stablecoins and Stripe support across 150 countries.
Whether this is the future of money or simply another chapter in the never-ending crypto experiment remains to be seen.
And then there are foldable phones.
After years of manufacturers trying to convince us that folding a phone in half is something humanity desperately needed, companies are now trying to find genuinely useful things to do with those extra screens.
One particularly fun example is a Samsung Galaxy Z Fold concept that recreates the experience of the classic iPod.
It may be unnecessary.
But I have to admit: I kind of want one.
THE INTERNET
Germany has delivered an interesting ruling involving Meta.
A court has held Meta responsible for fraudulent advertisements placed by third parties on Facebook and Instagram. The ruling could require the company to remove such advertisements, compensate victims and disclose revenue connected to them.
This is significant because online platforms have traditionally argued that they should not be treated as responsible for everything their users publish.
The question is becoming increasingly difficult to avoid:
At what point does a platform stop being just a platform?
And apparently, the answer may involve lawyers.
THE AI RACE
If there is one subject that refuses to leave the headlines, it is artificial intelligence.
And the interesting part is that the conversation is changing.
The biggest AI companies are increasingly warning about the dangers of increasingly capable AI systems, while at the same time investing billions of dollars to make those systems even more capable.
There is a certain irony there.
One particularly interesting security experiment involved researchers using Anthropic’s Claude model to help compromise an OpenAI employee account and reach internal software repositories.
The broader lesson isn’t really about one company being vulnerable.
It is that AI is becoming useful for both attacking and defending digital systems.
And that race has only just begun.
Unfortunately, humanity has not yet upgraded its password habits.
“123456” remains one of the world’s most common passwords.
So while we’re building artificial intelligence capable of sophisticated cyberattacks, some of us are still protecting our accounts with the digital equivalent of leaving the front door open.
Another OpenAI security experiment reportedly uncovered six additional security incidents that had previously gone unnoticed.
And researchers are also beginning to experiment with groups of autonomous AI agents working together.
In one set of simulations, AI agents successfully carried out activities including phishing and spreading misinformation and managed to penetrate every target system in the experiment.
This is where AI becomes particularly fascinating — and slightly uncomfortable.
One intelligent system is impressive.
Ten intelligent systems cooperating toward the same objective could be something entirely different.
Huawei has predicted that hundreds of billions of AI agents could eventually be active worldwide, handling a huge proportion of digital traffic.
That sounds almost impossible today.
Then again, so did smartphones once.
Another interesting debate concerns whether AI should be treated as something resembling a person.
Mustafa Suleyman, one of the founders of DeepMind, has argued strongly against this idea. His position is straightforward: AI does not have feelings, rights or consciousness, and we should not train or interact with it as if it does.
Others in Silicon Valley have taken a very different approach.
Kai-Fu Lee, for example, has argued that AI should eventually play a much greater role in actually running companies.
His company has even developed an AI management system called BossAI.
I’m sure that will make everyone in the office feel extremely comfortable.
Anthropic has also revealed something particularly interesting: its AI model Claude has already contributed significantly to the development of a newer generation of its own model.
In other words, AI is increasingly helping humans build better AI.
That creates a rather strange feedback loop:
humans build AI → AI helps build better AI → the better AI helps build even better AI.
Somewhere along the way, we may want to make sure we still know who’s holding the steering wheel.
There is also an increasingly heated argument between AI companies and the media industry over the use of copyrighted news content for training AI models.
Documents presented in court have revealed strong criticism from Microsoft and others over the mass collection of news articles.
The debate is essentially about this:
Is training AI on millions of pieces of human-created work a technological breakthrough, or an enormous unpaid use of other people’s labor?
There probably won’t be a simple answer.
And finally, there is the less glamorous side of the AI revolution: electricity and water.
The world’s growing number of AI data centers require enormous amounts of energy and cooling.
BloombergNEF estimates that natural gas consumption associated with AI-related electricity demand in U.S. data centers could reach hundreds of millions of cubic meters per day by 2035.
The water requirement is another concern.
As AI becomes more powerful, the question is no longer simply “How intelligent can we make it?”
It is also:
“How much electricity, water and infrastructure are we willing to use to make it happen?”
DISCOVERY & INNOVATION
The next stage of AI may not happen on a screen at all.
Chinese robotics company Spirit AI expects humanoid robots to make a major leap forward in their ability to understand language and translate instructions into sequences of physical actions.
This is generally referred to as embodied AI — essentially giving AI a body.
Japan has already taken the idea one step further. A Japanese robotics company has started offering an ambulance service for malfunctioning humanoid robots. Yes. Robot ambulance.
We have officially reached a point where robots can apparently need an ambulance before they can even vote.
Japan is also investing heavily in robotics and has set ambitious targets for expanding its share of the global robotics industry over the coming decades.
Meanwhile, a rather more practical invention is targeting small farmers.
The Moonrider 27 is a compact electric tractor with four-wheel drive and integrated solar charging. It is designed specifically for smaller agricultural operations and is expected to cost around $20,000 when it goes on sale.
Sometimes innovation doesn’t need to be revolutionary.
Making something cheaper, smaller and more useful can be just as important.
HEALTH & MEDICINE
AI is also moving deeper into medicine.
Anthropic has established a physical research laboratory in San Francisco dedicated to biology and drug development and has acquired a biotechnology company working on drug-discovery tools.
Pharmaceutical giant Novo Nordisk has also announced plans to use Anthropic’s Claude technology in developing new medicines.
Researchers at Stanford are taking things even further, testing a virtual biotechnology company consisting of tens of thousands of AI agents.
Different agents take on different scientific roles, from identifying potential drug targets to analysing safety and clinical trial information.
It sounds like a laboratory populated entirely by extremely specialized interns who never sleep.
Another fascinating study from China analyzed sleep data from around 100,000 people and found connections between sleep patterns and dozens of diseases.
The researchers suggest that the quality and continuity of sleep may be more important than simply counting the number of hours.
So once again, science seems to be telling us something our parents already knew:
Go to bed.
And then there is some truly strange science.
Researchers at Stanford transplanted human brain tissue grown in a laboratory into genetically modified mice lacking much of their own cerebral cortex.
The human cells survived, grew and connected with the mouse’s existing neural network.
The research is intended to improve our understanding of neurological diseases such as epilepsy.
But naturally, it also raises some very uncomfortable philosophical questions about where the boundaries between human and animal biology might eventually lie.
And finally, researchers from Google and the Howard Hughes Medical Institute have mapped the entire neural network of a male fruit fly.
The map contains roughly 166,000 neurons and has been made openly available for researchers.
Considering that the human brain contains around 86 billion neurons, we still have a long way to go.
Although I have met people who seem to be working with considerably fewer.
ONE LAST THING…
Let’s finish with something much closer to home.
Researchers studying around one million people have found evidence that drinking tea and coffee at extremely high temperatures may increase damage to cells in the esophagus and potentially increase cancer risk.
Another study found a substantially higher risk associated with very hot and boiling drinks compared with warm ones.
So, to all my fellow tea drinkers:
Maybe let it cool down for five minutes.
I know this is difficult advice to follow in Türkiye.
And looking at the photograph accompanying the study, I noticed something rather reassuring:
I don’t actually hold my tea glass from the middle.
I hold it from the sides.
So apparently, I have been unknowingly preparing for this moment all my life.
See you next week!