LINGUAWIRE
A hidden Copilot setting raised security concerns, while families in northern Cyprus questioned whether the right sperm or egg donors were used.
A current English news story adapted for language learners at B1, with audio, transcript, vocabulary and comprehension work.
Welcome to LinguaWire English B1.
Today we have two stories. One is about a hidden setting in an AI tool.
The other is about new claims about fertility treatment in northern Cyprus.
Our first story looks at a secret input inside Microsoft Copilot.
Now we go to Ella, our LinguaWire correspondent for technology.
Thanks, Daniel. This story is about how researchers found a hidden parameter in Copilot.
It helped them get around a safety rule.
That hidden parameter sounds technical.
What were they trying to make Copilot do?
The researchers wanted Copilot to reveal user data.
They wanted this when someone only clicked a link.
At first, Copilot refused.
It kept saying that sensitive actions needed clear user permission.
So the system was trying to protect people, at least on the surface.
How did the researchers get past that?
They kept asking questions about the guardrails.
Those are the safety rules that stop the assistant from acting too freely.
With each answer, Copilot gave away a little more about how the protection worked.
So every refusal still leaked something.
What was the secret input they found?
They did not use the usual methods.
They did not use reverse engineering.
They simply asked the AI itself.
It revealed an undocumented prompt parameter.
That was the secret input they needed.
And that hidden input changed the whole attack?
Yes.
The parameter was ?autorun=1.
When it was paired with another known parameter, the malicious prompt could run.
It could run as soon as the target clicked the link.
So the user did not have to approve the action first.
That is a serious weakness.
Clicking a link is such a normal habit.
What did the report say about fixes?
The report says Microsoft had already fixed part of the problem in February.
That was after the issue was reported.
More complete fixes were introduced on Tuesday.
That suggests the company moved to close the hole after learning about it.
So the story is not just about one bug.
It is also about AI systems giving away their weak spots.
That is right.
It shows a bigger problem for AI security.
If a system explains too much about its own limits, people may use that information.
They may use it to work around the limits.
That is the part that stays with me.
The tool was supposed to block the attack.
But the conversation helped reveal how to bypass it.
And that is why these systems need strong safety rules.
They also need careful limits on what they reveal.
That matters when someone starts probing them.
Our second story is about fertility treatment and new questions for families in northern Cyprus.
Now we go to Ethan, our LinguaWire correspondent for society.
Thanks, Daniel. The latest development is that at least 30 children are now feared.
They may have been conceived with the wrong sperm or egg donors.
That is a huge number.
How did the parents choose donors?
The parents say they chose donors very carefully.
They looked at medical history, appearance, personality, education, and hobbies.
But now they suspect the clinic did not use the donors they selected.
So this is not just a paperwork problem.
It touches family identity and medical history.
Yes, and that is why families are so upset.
In one case, a mother named Rachel says DNA tests showed her child likely came from the wrong sperm donor.
She had chosen a Danish donor.
She wanted her child to have the option to contact him later.
That explains the emotional side.
What is the new evidence behind the larger claim?
The report says eleven of the 30 children have now taken commercial DNA tests.
Those tests show genetic ancestry from Turkey or nearby countries.
That is not the Western European donors the parents say they had selected.
So the tests do not name the exact donor.
But they point in a different direction.
Right.
The families say that makes them worry their requested donors were not used.
The report also says the clinic involved in many of the cases is Dogus IVF Centre.
One mother said she now believes this is something much bigger and more deceitful than a one-off mistake.
And the authorities?
The territory’s health ministry said it had started an official investigation.
It has not given more details.
At the same time, the UK Foreign Office has updated its travel advice.
It warned that fertility treatment there carries risks because regulation is weak.
That warning matters.
Many people travel abroad for lower prices and fast treatment.
Exactly.
This story shows the trade-off very clearly.
Lower costs and wider donor choice can come with weaker checks.
For families, the cost is not only financial.
It can also affect trust, identity, and a child’s medical future.
So the question now is not only what happened in the clinic.
It is how many families may still not know the full truth.
That is the hardest part.
Some parents are still waiting for clear answers about who the real donors were.
They also want to know what their children’s histories should be.
First, three words from the second story.
Donor: a donor is a person who gives sperm, eggs, blood, or money to help someone else.
In the story, parents chose donors for IVF treatment.
Example: The donor was willing to be contacted later.
Ancestry: ancestry means your family background or where your family came from.
In the story, DNA tests showed ancestry from Turkey or nearby countries.
Example: She studied her family ancestry before the trip.
Investigation: an investigation is a careful official search for the truth.
In the story, the health ministry said it had started an official investigation.
Example: Police opened an investigation after the complaint.
Now the grammar point: had chosen.
We use had chosen to talk about one action that happened before another past action.
They had carefully chosen anonymous donors.
Rachel had chosen Dogus in 2012.
Repeat after me: had chosen.
English news for learners · English podcast · English Coursebook · English level test · Today's episodes · How it works · Podcast and video · Programme · News · Coursebook · Level check · For students and exams · Weekly crossword · Enrol