Google’s universal AI assistant vision, fighting spam and backing up photos
Google plays its cards early in 2025, with this year’s I/O more than delivering on expectations.
This is when I truly note the onset of summer. Never mind, that's just me being me. However, the annual developer conferences, by the big tech, are a big deal. May and June are the months when Google, Microsoft, Apple, Meta…pretty much all the important names you can think of, will lay down their vision with all things software, apps, developer engagement, and now increasingly, all things artificial intelligence (AI). Google plays its cards early in 2025, with this year’s I/O more than delivering on expectations. And then some. I had mentioned to a friend recently, Google has rewritten some chapters, and reset some benchmarks. Having the first mover advantage this summer has most certainly worked in terms of timing, and gives competition little time to realign. If at all they need to.
AI announcements, rightly so, needed as much time as possible in the keynote. There was a lot to pack in. I’ll summarise this for you.
- Google is intent on morphing Gemini as we know it, into what they call a “universal AI assistant”. Some call them “AI agents” or “Agentic AI”, Google chooses to call this a universal AI agent. Key to this will be AI’s ability to use world knowledge, reasoning, and simulate natural environments, just as a human brain would do.
- “Our recent updates to Gemini are critical steps towards unlocking our vision for a universal AI assistant, one that's helpful in your everyday life, that's intelligent and understands the context you're in, and that can plan and take actions on your behalf across any device. This is our ultimate goal for the Gemini app, an AI that's personal, proactive and powerful,” notes Demis Hassabis, CEO of Google DeepMind, in a session of which HT was a part.
- Anthropic, last year, introduced MCP an open-source standard. It is also called the “USB-C port of AI”. Simplicity and wide-spread support is key, since app developers can use MCP to enable their apps or agents to talk to other apps and services.
- Google is rolling out significant upgrades for the Gemini 2.5 Flash and Gemini 2.5 Pro models. Incoming improvements to the Gemini 2.5 Pro model, add new reasoning capabilities with Deep Think mode. Its specific focus on complex math and coding tasks, will be relevant for Gemini’s march towards an ‘agentic AI’ vision.
- The lighter Gemini 2.5 Flash receives improved reasoning, multimodality, code and long context. For now, the updated 2.5 Flash is available as ‘experimental’ in Google AI Studio for developers, in Vertex AI for enterprises, and the Gemini app for everyone — its final release is pegged for early June.
- Google’s newest generative media models, are arriving now. The video generation model Veo 3 and the image generation model Imagen 4, find new capabilities. Alongside new updates for the previous generation Veo 2 model, including camera controls.
Google is introducing two new AI subscription plans, and this shouldn’t be a surprise, since there is pressure on the widening AI tools, to generate revenue for the tech giant. There is the Google Al Pro (this is a renaming of the current Google AI Premium plan, with some add-ons), and Google Al Ultra that will cost $249.99 per month.
Google Meet’s logical successor? The 3D communication platform,, as Google Beam is being called, uses an AI volumetric video model is what makes these calls appear fully 3D from any perspective. It transforms standard 2D video streams into realistic 3D experiences.
- Android 16, the final release expected in a few months, will introduce the biggest visual changes to Android in more than four years. On the agenda now is ‘Material 3 Expressive’, with focus on fluidity of motion, aesthetic updates and a revamped interface. It had been a while — Google last gave Android a fresh coat of paint in the summer of 2021, with the Material You design language.
- Gemini AI will now find a home in Wear OS smartwatches, Google TV based smart TVs, and in car with Android Auto as well as those with Google built-in. These integrations roll out in due course.
- In the coming months, Android phones will add scam detection in Google Messages as well as in-call scam protection against methods such as screen sharing. The actions that’ll be blocked during a call include disabling Google Play Protect, side-loading an app on the phone, or granting accessibility permissions which can give a newly downloaded malicious app access to user’s device and steal sensitive data such as banking details.
- Google recently launched an AI based Scam Detection in Google Messages, and Phone by Google. This on-device AI processing should be able to detect conversational scams, the kind that usually sound very innocent at first, but the underlying malicious intent is designed to steal money from unsuspecting individuals. When Scam Detection discovers a suspicious conversation pattern, it warns the user in real-time.
- Google is also integrating something called a Key Verifier, which will verify the identity of the other party through public encryption keys, protecting end-to-end encrypted messages in Google Messages.
- An Android update for existing phones, alongside the Android 16 rollout, will further reinforce Factory Reset protections, which will restrict all functionalities on devices that are reset without the owner’s authorisation. The owner of the stolen phone will also have more control via the Remote Lock feature, and added to the mix will be a security challenge question, helping to prevent unauthorised actions that a thief may be attempting.
Last week on Wired Wisdom: Android’s security envelope, an impressive 'Tank' backpack, and OpenAI’s corrective steps
TECH SPOTLIGHT: Nothing CMF Phone 2 Pro
It isn’t often that mid-range Android phones come this good. Nothing, the UK-based tech startup, is never short as far as that standout element is concerned. That’s perhaps best represented by their flagship phones, but the more affordable CMF line-up. Not often do you get a tiny screwdriver as part of the phone’s package — in this case, that’s to get the back cover accessory in place, if that’s what you want. I preferred to use the CMF Phone 2 Pro as is, because the green colour way looks really good. Plus, there’s a slimness to the overall design, which many of you may want to leave unaltered. I want to talk about the accessories for a bit, because that’s something none of the CMF Phone 2’s rivals have. It is a full combination that you can optionally opt for, either in totality or in pieces.
The universal cover can be installed by removing the screws already on the phone, placing this cover and reattaching the screws. The closest we’ll get to building your own phone. Which reminds me, where are all those modular phones which everyone raved about, before they got busy with AI? There’s the interchangeable lens combination, for which you will need the universal cover since it provides the holding mechanism for these lenses. The fisheye and the macro options should work well if you tend to use the phone for specific types of photos. I must note that even without, the CMF Phone 2 Pro’s camera is very, very capable. Even more so, with the latest Nothing OS update. The lanyard is self explanatory, and there is an audience for it (you may not be, ignore and move on). The wallet and stand accessory has genuine utility, as either. It attaches to the universal cover, and if you aren’t worried about losing your credit cards if your phone gets snatched or stolen, be my guest. But I’d end up using the stand option more, because that simply works brilliantly on the workstation and the bedside table. Except when the phone’s to be plugged in for charging.
BUILD A FIREWALL
Cellular service providers have to take the lead in the battle against spam and scams. There are no two ways around that uneasy conversation. Apps such as Truecaller, as brilliant as they are, aren’t on everyone’s phone yet. And even if they are, network-level protective mechanisms are really the only way to counter the volume of scam and spam floating around on our mobile networks, and ingenuity that’s being shown by these scammers. They could’ve used their brain cells towards nation building, but targeting vulnerable senior citizens and swindling them of their savings, is the path they chose instead. But I digress.
Airtel’s spam detection for incoming calls and incoming SMS on their network, has been a responding success. A much-needed warning for ‘suspected scam’, enough to alert even the most unsuspecting and technologically not inclined, phone user. Bharti Airtel isn’t done yet. They’re now integrating something called ‘Fraud Detection Solution’, which they say is a first of its kind solution worldwide. It may well be, because the scope and relevance, at least with the potential that’s being talked about, is astounding.
This solution will use artificial intelligence (AI; of course, would you expect anything that doesn’t have AI these days?) detect and block malicious web links that are often included in messages sent by spammers and scammers in WhatsApp, Telegram, Facebook, Instagram, SMSes, and they also say across emails and web browsers too, in real time.
“Our AI based tool scans internet traffic, checks with global repositories and our own database of threat actors in real-time and blocks fraudulent websites,” notes Gopal Vittal, Vice Chairman and Managing Director, Bharti Airtel, adding, “Our solution has already reached a remarkable level of accuracy in the 6 months of trials.” The service works at a network-level which means no app download or update is necessary, and will be auto- enabled for all Airtel mobile and broadband customers at no additional cost.
You could ask a question about how much data is Airtel getting access to, for every link that it detects and flags in any SMS or email or web browsing sessions? There are no clear answers yet, in terms of how the link data is collected, what else is alongside, and how secure is the chain of scan and storage. Airtel says that for now, this feature rolls out in the Haryana telecom circle, and a nationwide rollout will happen in the coming weeks. We really need more from mobile service providers to counter the risks that scammers and spammers pose. Perplexing enough, is Reliance Jio’s silence on any similar network-level features. It may perhaps be too much to ask Vodafone Idea, or Vi, considering they’ve more struggles to contend with.
WHAT’S ON MY MIND?
Finally, a viable alternative for Google Photos? Think about it, if not Google Photos for backing up your memories, then what? See, you’re struggling to find an answer. A solution may be at hand, with the Swiss privacy-focused tech company Proton, announcing a fairly significant update for its Proton Drive app. Across Android and iOS, as well as the web, Albums in Proton Drive, makes it a viable alternative to that combination of Google Photos and Google Drive. Users can organise photos that have already been backed up further into specific albums, create a shared album, and mark favourite photos within the gallery. Arriving soon, ability to generate links to share photos via these links, instead of giving a friend or family member access to an entire album. I’d say worth considering, if you want better encryption for your photos. By better, I’m talking about choosing between more or less secure methods. After all, Google has a bunch of AI features in Photos, for which it’ll need some level of access to data.
Other unmissable tech pieces, from the past week…
