With AI constantly dominating the news cycle, it might feel intimidating to keep up with the stories which are releasing. Whether it be ChatGPT getting a new update or Google Gemini being implemented into apps like Instagram, there are ever occurring changes happening in the world of artificial intelligence.
We’ve rounded up the five key stories within the world of AI that you should know about. Regardless of if you’re a novice in the world of AI and just want to have a gander at the news or you’re a well seasoned AI veteran and need some more information about the changes occurring, we’ve got you covered.
Does ChatGPT ignore reliability concerns?
Many people, particularly students, use ChatGPT and other AI companions to seek and analyze information to support essays or academic articles. A study published in August 2025 shows that ChatGPT was not aware when articles had been retracted or were otherwise problematic meaning they could be obtaining information which was incorrect or posed other issues.
The study tested the 4o-mini version of ChatGPT and the results showcases that the AI wasn’t aware of when an article had been retracted or at least didn’t take this into account when asked for relevant information about the articles or when asked to fact check a claim made by them.
In summary, this shows that users should be cautious of summaries of academic information made by AI models. It is worth noting that ChatGPT states underneath the text input box that ‘ChatGPT can make mistakes. Check important info.’ so be sure to fact check information that it may spit out at you where possible.
OpenAI releases upgrade for ChatGPT coding functions
A big upgrade to the ChatGPT codex for agentic coding now makes the AI much better for helping and editing code. The changes include new VS Code extension, sync support between web and terminal and a whole load more features.
Codex now works in your local Windows terminal or IDE, with OpenAI stating that your ChatGPT account connects your local code with Codex on web so you can work seamlessly between platforms without losing state.
ChatGPT has a ‘secret’ quiz feature
OpenAI released a new feature called QuizGPT which you can only access by asking for it specifically. This generates flashcards for you and can ask you questions, making it extremely helpful for studying.
You can access the feature by using the following prompt: “Quiz me on [TOPIC/SUBJECT] in QuizGPT as flashcards.”
ChatGPT will then switch to QuizGPT mode and create flashcards for whatever topic you choose. You can provide additional context to ensure that the flashcards suit your needs like clarifying if you’re a beginner or expert in the topic.
Meta might use Gemini to boost Instagram and Whatsapp AI features
According to The Information, a Meta spokesperson has stated that they may start utilising Google’s Gemini AI to enhance user experience within WhatsApp and Instagram. The company is exploring the possibility of adding Gemini into the Meta AI chatbot to handle conversational and text-based search queries.
The spokesperson stated that the company will have an “all-of-the-above approach” to AI development, whether its external partnerships or internal innovation. The company implementing external AI companions means that they can focus on refining their own technology rather than scrambling to release something to keep in line with the competition.
OpenAI confirms police access to ChatGPT in extreme cases
New privacy concerns have been struck up after OpenAI admits that conversations which are flagged for violence could be reviewed by humans or even shared with the police. This means that the confidentially promised to users is not longer valid, as many people believed that their AI conversations were unable to be seen by human eyes.
According to OpenAI, the process starts when its automated system detects a possible threat, which then escalates the messages to a review team. The human reviewers can then conclude if the risk is immediate and then OpenAI will alert authorities.
While this could be a great thing for ensuring safety for users, this also could be seen to cross a line regarding user trust meaning OpenAI now have to play a balancing act.
