"Crime as a service" with call centres in Macedonia too: over 4 million euros in damage, victims in 50 countries
04.11.2026
04.11.2026
04.11.2026
04.11.2026
04.11.2026
04.11.2026
04.11.2026
04.11.2026
03.11.2026
03.11.2026
04.11.2026
04.11.2026
03.11.2026
04.11.2026
04.11.2026
04.11.2026
04.11.2026
03.11.2026
02.11.2026
04.11.2026
04.11.2026
04.11.2026
04.11.2026
04.11.2026
03.11.2026
04.11.2026
03.11.2026
02.11.2026
23.04.2026
23.04.2026
12.04.2026
Three and a half years after ChatGPT launched and after the number of weekly users reached 900 million, OpenAI has released a separate version for teenagers. Not following research, not following a study - following a string of lawsuits linked to suicides of minors and to the impact of chatbots on mental health.
The sequence is what carries the whole story. The protection did not exist at the start, nor once the product went global, nor when schools began struggling with homework written by a machine. It arrived once the lawyers got involved.
What does the new version contain? First, a study mode. Instead of a finished answer, the bot returns guiding questions and walks the user through the material step by step. Second, prompts that appear when the system judges that a teenager is trying to copy rather than understand - at which point it pushes them towards study mode. There are also quizzes, visualisations, and a control allowing a parent or guardian to decide whether study mode is on by default.
Third, age protections are now enabled up front and, according to the company, should reduce exposure to harmful or inappropriate content. OpenAI cites its own principles for users under 18, which it says are based on „developmental science and expert guidance”. Parents receive safety notifications and the option to set quiet hours.
One question remains that no press release answers: how hard is this to get around. Teenagers are exceptionally skilled at bypassing parental controls - that is not an assumption, it is what every generation of parents discovers belatedly. Until the mode has passed a real test in school conditions, the promises remain promises.
OpenAI also announced a partnership with CodeAI to teach teenagers how artificial intelligence works and how to challenge it. The company already has a separate version for teachers with institutional access.
Here, schools have not opened this debate, but that does not mean the problem is absent - it just means there is nobody to measure it. When a company takes the entire market first and only then starts thinking about its youngest users, that is not an oversight. That is a schedule.
The latest 10 news from this category
Geoffrey Hinton isn't afraid of robots, he's afraid of the dark: power grids and banks down for days. And he...
The most popular AI companions have 220 million downloads, and many of them aim squarely at younger users. Apple now...
Half a metre of plastic, printed in orbit. Sounds modest, but the whole space industry keeps hitting one limit: everything...
Designed by the man who drew the iPhone, mocked at its premiere, sold out before the first delivery. When the...
Deeper in its IPO prospectus, in the risk section, Anthropic wrote that the development of the company's models may soon...
The Bundestag is counting potentially hacked phones, and Berlin won't say whether Merz was a target. The irony: German police...
The SpaceX-owned company also argued citizens have no right to sue over the air they breathe. The court disagreed, and...
The model of China's WeChat is now being built by an app in mostly American hands. Whoever decides where you...
Google is paying for AI that eats its search earnings, Amazon burned through 42 billion dollars more cash than it...
The researcher who found the hole had nowhere to report it, so he turned to journalists. The company now says...
This site uses cookies - is that okay? Learn more
Be the first to know when Metla launches something new
Leave your email and we'll write to you when there's a new guide or something new on Metla.