You are most likely living under a rock if you haven't been seeing the AI buzz all over the internet, especially with the fear that AI is coming for jobs. And honestly, it is only just starting.

Jack Dorsey cut half of Block's workforce, not because the company was struggling, but because AI made it possible. OpenAI crossed 900 million users and raised $110 billion. Google launched a real-time 4K image model. And Anthropic caught Chinese labs using 24,000 fake accounts to steal from Claude.

It was that kind of week. We broke it all down below for you incase you didn’t have time to dig through during the week.

ChatGPT just hit 900 million weekly active users. OpenAI announced the milestone on Friday, alongside news that the platform now has 50 million paying subscribers, a number the company says is growing faster than ever.

According to OpenAI, January and February are on track to be the biggest months for new subscribers in the company's history. That kind of growth in what is typically a slow quarter says a lot about where user appetite for AI tools currently sits.

The 900 million figure is up from 800 million reported back in October 2025, that is 100 million new weekly users added in roughly four months. OpenAI says the growth is not just in numbers either. As more people use the product, they claim it gets faster, more reliable, and more consistent for everyone.

OpenAI dropped these numbers alongside another major announcement, a $110 billion funding raise, one of the largest private funding rounds ever recorded. Amazon put in $50 billion, with Nvidia and SoftBank contributing $30 billion each. The round values OpenAI at $730 billion before the money comes in, and it is still open with more investors expected to join.

Anthropic has accused three Chinese AI companies of systematically extracting knowledge from its Claude model using fake accounts. The companies named are DeepSeek, Moonshot AI, and MiniMax. According to Anthropic, the three labs created over 24,000 fraudulent accounts and used them to generate more than 16 million conversations with Claude, all with the goal of making their own models smarter.

The technique they used is called distillation. It is a common and legitimate training method where a company takes a large AI model and uses its outputs to train a smaller, cheaper version of their own. The problem here is that these labs were using it on someone else's model without permission, essentially copying the work that Anthropic spent enormous resources building. Anthropic says the attacks specifically targeted Claude's most advanced capabilities including reasoning, coding, and tool use.

The scale of each operation varied. DeepSeek ran over 150,000 exchanges focused on foundational logic. Moonshot AI had more than 3.4 million exchanges targeting reasoning, coding, and computer vision. MiniMax was the most aggressive, racking up 13 million exchanges, and Anthropic says it watched in real time as MiniMax redirected nearly half its traffic to siphon from Claude's newest model the moment it launched.

Beyond the competitive threat, Anthropic is raising a national security concern. The company says its models are built with safeguards to prevent things like bioweapon development or cyberattacks. Models trained through distillation, it argues, are unlikely to carry those protections over, meaning dangerous capabilities could spread without the guardrails.

The accusations also land right in the middle of a policy debate about whether the US should restrict AI chip exports to China. Anthropic's position is clear, llimiting chip access limits the scale at which these kinds of attacks can happen. The Trump administration recently loosened some of those restrictions, allowing Nvidia to sell advanced chips to China, a move critics say undermines efforts to slow China's AI progress.

Block, the fintech company behind Square, Cash App, and Afterpay, has laid off more than 4,000 employees. The company, founded by Jack Dorsey, has gone from over 10,000 workers to just under 6,000 in one day. What makes this story different from most layoff announcements is the reason behind it. Block is not in trouble. The company openly says its gross profit is growing, it is serving more customers than ever, and profitability is improving.

Dorsey's explanation, shared in a lengthy memo posted on X, is that AI tools have fundamentally changed what it takes to run a company. He described seeing smaller, flatter teams achieve things that would have previously required far larger headcounts, and said that shift is accelerating rapidly. Rather than carry out slow, repeated rounds of cuts over months or years, which he believes damage morale and erode trust, he chose to make one clear and decisive move.

The severance package for affected employees is notably generous. Workers are receiving 20 weeks of base salary plus an extra week for every year of tenure, six months of healthcare coverage, equity vesting through the end of May, and $5,000 in transition support. Corporate devices are also being kept by departing employees.

Wall Street liked what it heard. Block's stock jumped more than 24% in after-hours trading following the announcement. Investors read the move as a signal that the company is leaning into efficiency and betting on AI to do more with less.

The reaction beyond the markets has been more divided. Some see Dorsey's decision as an early and honest acknowledgment of where the corporate world is heading. Others raise harder questions about what happens when large-scale job displacement becomes a pattern across industries, especially when the companies doing it are profitable.

Google has launched Nano Banana 2, an upgrade to its image generation model that the company is positioning as its most capable yet. Also known as Gemini 3.1 Flash Image, the model is set to become the default image generator across the Gemini app and other Google platforms. The idea is to bring what was previously only available in Google's more premium tools to a much wider audience, at a significantly faster speed.

The upgrade comes with a meaningful set of new features. Nano Banana 2 can pull from real-time web search data to generate images that are more accurate and context-aware, which is useful for things like education, localized marketing, and travel content. The model also supports high-resolution upscaling up to 2K and 4K, improved text rendering within images, and multiple aspect ratios including standard widescreen and vertical formats. Every image produced carries a SynthID watermark so viewers can always tell when AI was involved in creating it.

For businesses and creators, Google has also integrated C2PA Content Credentials into the model, a transparency standard that shows not just whether AI was used, but exactly how it was applied. On top of that, Nano Banana 2 is being rolled into Adobe Firefly, deepening Google's existing partnership with Adobe and giving creative teams a more seamless path from idea to production-ready visual.

The rollout is happening across Google Search, Google Lens, and Google's AI development tools, with enterprise access coming through Vertex AI. The original Nano Banana brought 13 million first-time users to the Gemini app within four days of launch and generated over five billion images within weeks. Google is hoping the sequel does even more, and it is arriving at a moment when the competition in AI image generation, particularly with OpenAI, is as intense as it has ever been.

Democratize and productionize Gen AI across your entire org with Portkey's suite of AI gateway, observability, guardrails, and prompt management modules.