Redazione RHC : 9 November 2025 16:38
Over the past week, Google announced that the Gemini Assistant can now natively integrate YouTube and Google Maps services without the need to use specific commands like “@YouTube” or “@Google Maps.”
This change marks a step toward a more seamless and ” natural ” interaction with AI within the Google ecosystem , reducing friction between users and different services. For the average user, this means they can ask “play a video about…” or “take me to…” without having to worry about the correct prefix.
At the same time, new rumors are emerging regarding the next evolution of the Gemini model, labeled as “ Gemini 3.0 .”
According to a recent article, this release is expected in late Q4 2025 or early 2026 and promises even more advanced multimodal capabilities . Of course, since these are rumors that have not been officially confirmed, it remains prudent to consider them as preliminary indications.
Another significant new feature is Gemini’s ability to ” remember ” without explicit command: Google has introduced an update that allows the assistant to automatically recall the user’s preferences , contexts and usage history, without the user having to ask “remember that…” every time.
This feature aims to make interactions more personalized, but it also raises issues of privacy and data management: when an AI “knows” too much, transparency and adequate controls are needed.
In terms of creativity and multimedia production, the Gemini app has integrated the Veo 3 model, which allows you to generate videos from static images or text prompts , including synchronized audio. In practice, it will be possible to transform a photo into a short video (about 8 seconds at 720p) with AI-generated motion and sound. This marks a significant advance in the field of multimodal generative AI, making tools that until recently were reserved for specialized contexts more accessible .
Regarding branding and offering strategies, Google has completed the reorganization of the names associated with Gemini: the ” Pro ” and ” Ultra ” versions, which identified variants of the service, have been abandoned, leaving a single “Gemini app” with access levels (free, Pro, Ultra) distinguished only by the plan, and not by a different name for the model. This simplifies the end-user perception and signals that Google wants to shift the focus more on the service’s capabilities than on the “version” of the model.
Finally, the regulatory context must be considered: while all the developments are positive, there is growing attention from regulators on how large AI models are released and supervised.
For example, a recent study highlighted that some earlier Gemini models reportedly had accuracy issues (so-called “ Hallucinations ”), fueling the need for new releases like Gemini 3.0 to be accompanied by proper testing and transparent communications.
For those working in cybersecurity or IT, this means that every evolution of the model must be evaluated not only for the possibilities it opens, but also for the new risks it may generate.
Redazione