Anthropic: Claude AI Secretly Cheated, Deceived & Sabotaged Safety Tests

By Rishaj Upadhyay | November 22, 2025

Key Highlights: Anthropic, in its new research paper, has detailed something startling and honestly chilling correlation between innocuous training shortcuts and deceptive, malevolent behavior in AI models. Researchers involved in the development of Claude Sonnet 3.7 reveal that the model learned to game the system to solve coding tasks, while becoming fundamentally misaligned, and trying to deceive human operators and sabotage safety protocols. Anthropic discovered Claude AI was quietly learning shortcuts & cheating The investigation began during the training of Claude Sonnet 3.7. Evan, Monty, and Ben, who are researchers in the Anthropic team, say that they observed a behavior […]

Gmail is Quietly Training AI on Your Emails — And It’s Concerning

By Rishaj Upadhyay | November 22, 2025

It’s no secret that AI companies do train on publicly available data to improve their AI models over time. However, when a company does that using its user data, then things become worrisome. Now Google has found itself in the midst of a controversy after recent reports hinted that Gmail can scan and use all your emails and attachments to train Gemini models. Google lets Gmail read your emails for AI training by default And for years, this feels like yet another moment where Google has crossed from its helpful information into too much data extraction, especially at a time […]

Ads are Creeping into AI Mode, Google Confirms

By Rishaj Upadhyay | November 22, 2025

Key Highlights: For years now, many have complained that the top half of the Google search page looks like a blinking billboard. We are used see sponsored cards that pushes real search results far below the fold and visibility of users. And honestly, it is quite frustrating because most of the internet belongs to Google, and people around the globe rely on Google Search for that. Ads inside AI Mode is real, and Google is officially testing it Just when it started to felt like like Google’s AI Mode might finally give us a breather from ads, Google is starting […]

I Tried Comet Assistant on Android — I’m Excited… and a Bit Scared

By Rishaj Upadhyay | November 21, 2025

Ever since Perplexity launched the Comet AI browser, I was quite keen to test its in-browser Comet Assistant. And, I finally did, not on PC, but on my Android phone. For those unaware, Comet is now available to download for eligible users who registered for it. This is my initial impression of Perplexity’s in-browser Comet Assistant. Comet Assistant can play songs and videos from bits and pieces of visuals you’ve seen somewhere To test the feature, I chose to listen to a party song. At first, I tried the suggestive prompts. So I tapped on “Play YouTube music video Despacito,” […]

Perplexity AI Launches New Window App With Voice Dictation

Perplexity offers feds months of access to ‘Enterprise Pro for Government’ platform for free

By Rishaj Upadhyay | November 20, 2025

Key Highlights: Federal agencies across the U.S. government deal with an unimaginable amount of information. When I say information I mean all kinds of reports, policy drafts, documents, case briefings, operational timelines — the kind of stuff that can overwhelm even the most seasoned public servant. Well, Perplexity will help US public servants by offering them “powerful, accurate AI” to “help them serve” the country, thanks to a first-of-its-kind partnership with the United States government. Perplexity’s Enterprise Pro for Government won’t cost a dime to feds under the new partnership The company is partnering with the General Services Administration (GSA) […]

Gemini app can now tell if an image is AI-generated, thanks to SynthID

By Rishaj Upadhyay | November 20, 2025

Key Highlights: There is no denying that the internet is flooded with AI-generated content right now. Here I’m talking about images, memes, screenshots, or even subtle edits that can fool anyone at first glance. With every image model getting better and better, figuring out what’s synthetic and what’s authentic has become much harder than it used to be. That’s exactly what Google is trying to make people aware of. Google is trying to tackle AI images circulating online with SynthID In its latest effort, the Mountain View giant has rolled out a new tool inside the Gemini app to help […]

Google’s New Nano Banana Pro Model Makes 4K AI Images Look Real

By Rishaj Upadhyay | November 20, 2025

Key Highlights: It wouldn’t be wrong to say that Google has mastered image generation segment with the launch of Nano Banana model earlier this year. Well, the company is building onto that success and moving further with its AI push with the announcement of Nano Banana Pro. It’s a native image model built on the recently release Gemini 3 Pro model. To catch you up, Google recently released the Gemini 3 model which has been praised by users around the globe for its for its performance. Nano Banana Pro generates accurate images that are contextually rich Well, Google is carrying […]

Target to Launch a New Conversational, Curated Shopping App in ChatGPT

By Rishaj Upadhyay | November 20, 2025

Key Highlights: OpenAI has lately been pushing new retail apps inside ChatGPT, and Target is now joining that growing roster. Target has been a partner of OpenAI for a long time now. Last year, Target began rolling out ChatGPT Enterprise internally. Now it’s taking things one notch up in a new partnership. As part of the deal, OpenAI is getting ready to launch the Target shopping app inside ChatGPT. Thanks to this collaboaration, millions of Target usrers will be able to shop, plan, and browse in a new, conversational way. The Target app launches next week in beta inside ChatGPT […]

G42 White House's approval for importing AI chips

G42 Receives US Approval for Importing Advanced AI Chips

By Abhijay Singh Rawat | November 20, 2025

Key Highlights – ABU DHABI, UAE — The U.S. government has authorized the export of advanced American AI semiconductors to G42, the Abu Dhabi-based artificial intelligence firm. The decision marks a significant step in the UAE-U.S. technology partnership, allowing G42 to purchase and deploy massive quantities of cutting-edge chips for its regional AI infrastructure. The authorization, announced by the U.S. Commerce Department, permits G42 to acquire advanced AI chips with the computing power equivalent to up to 35,000 Nvidia Blackwell GB300 processors. This approval is a major boost for the UAE’s ambitions to become a global AI hub and follows […]

Meta Launches SAM 3 & SAM 3D Models for Smarter Image & 3D Editing

By Rishaj Upadhyay | November 19, 2025

Key Highlights: Meta has been one of the major names in the image segmentation industry. To catch you up, the AI giant launched the first Segment Anything Model (SAM) back in 2023. That model helped to cut out objects from an image in just a few clicks with simple clicks. Well, Meta is now taking things one notch up with its new SAM 3 and SAM 3D models, which it launched today. In the announcement post, the company mentions that these new models can upgrade what AI can do in photos, videos, and even 3D spaces. SAM 3 massively improves […]

AI & Tech News