Google Earth Pulls the Plug on AI Image Generator: A Playground for Fake News?
Just about 48 hours after rolling out its snazzy new AI image generation feature, Google hit the brakes and suspended it on August 2nd, citing the need for beefed-up security. This comes after experts warned that the tool could let folks slap fictional scenes right on top of real satellite imagery, turning it into a potential megaphone for disinformation.
Here’s the gist, as reported by the BBC: The feature mashed Google’s image generator, Nano Banana 2, right into Google Earth. All you had to do was type in a prompt, and boom—you’d get a fictional scene layered onto authentic satellite views, aerial shots, or 3D maps. And yep, you could download and share these creations anywhere else on the web.
Google’s official line was that these AI-generated images come with invisible watermarks, and users could double-check them using Gemini or Google Lens. But here’s the kicker: BBC’s own testing found that some of those verification measures could be sidestepped, and third-party detection tools sometimes couldn’t tell a real image from a fake one either.
World’s Most Valuable Company Crown Shifts Again: Nvidia Back on Top
It was a wild Friday on Wall Street. Apple’s stock took a nosedive, dropping 7.35% and wiping out nearly $358 billion in market cap in a single day. Meanwhile, chip titan Nvidia saw its shares close up 2.93%, reclaiming the title of the world’s most valuable publicly traded company.
Big US Firms Are Trading in American AI for Chinese Models to Cut Costs
From AlphaGo dominating the Go board to ChatGPT taking the world by storm, the US has long held a serious lead in the large language model game. But lately, China’s AI upstarts are shaking up the status quo. According to reports from sources like NUPIAO, major outlets including the Wall Street Journal have highlighted that Chinese AI unicorn Moonshot AI’s new open-source model, Kimi K3, sent shockwaves through the capital markets. They’re comparing the buzz to the market panic triggered when DeepSeek launched back in 2025—and even Elon Musk chimed in, calling it “impressive.”
And it’s not just talk. Some major American companies are voting with their wallets. The Associated Press reports that crypto exchange Coinbase has confirmed it’s switching to Chinese AI models to trim expenses. Airbnb, meanwhile, has adopted Alibaba’s Qwen model, singing its praises as “fast and cheap.”
Amazon Officially Wraps Up Its Whopping $50 Billion Investment in OpenAI
On July 31st, Amazon dropped a bombshell in its quarterly 10-Q filing: it has fully completed its massive $50 billion total investment in OpenAI. Breaking it down, Amazon put $15 billion into OpenAI’s Series C preferred stock back in Q1, and pledged an additional $35 billion. In Q2, they funneled $13.7 billion of that commitment into more Series C preferred shares. Then, after the reporting period wrapped, they poured the remaining $21.3 billion into the company. Amazon also noted that if OpenAI goes through with an IPO or some other liquidity event down the road, those Series C preferred shares would convert into common stock. They expect to be subject to the usual lock-up period post-IPO, and will continue to be bound by applicable securities laws after that.
xFusion’s AI Lab Platform Drops Official DeepSeek-V4-Flash Version
On August 1st, xFusion announced that the official DeepSeek-V4-Flash-0731 model has landed on its AI Lab platform. Users can now access it via PC and take it for a spin online.
