Online publishers are seeing search referrals implode as Google swaps blue links for AI-generated summaries. The WSJ reports that features like AI Overviews and the new conversational AI Mode are siphoning organic traffic — Business Insider slumped 55% since 2022, HuffPost and The Washington Post face half their search traffic lost — forcing layoffs and strategy pivots. Google claims AI tools boost engagement and still promote links, but publishers say the shift turns Google into an “answer engine,” threatening journalism and prompting deals or lawsuits over content rights

Mary Meeker, known as the “Queen of the Internet,” released a report highlighting the unprecedented pace of artificial intelligence adoption. The study shows AI is being invested in, developed, and used faster than any previous technological change. ChatGPT reached 100 million users in under two months, far quicker than Facebook’s 4.5 years. Meeker attributes this rapid growth to aggressive development by tech companies, decreased costs, and widespread buy-in from businesses. The report also notes increased mentions of AI in corporate earnings reports and a 167% year-over-year increase in new AI models since 2020.

ChatGPT, the wildly popular AI chatbot, faced a significant outage on Tuesday, leaving users frustrated and unable to access its conversational capabilities. Over 1,500 reports flooded Downdetector, as the platform displayed error messages and failed to respond to queries. The disruption comes on the heels of OpenAI’s announcement of reaching $10 billion in annual recurring revenue, a staggering leap from last year’s $5.5 billion. Despite the impressive growth, the company reported a substantial loss of approximately $5 billion in the previous year. The outage coincides with the recent release of Advanced Voice, an update for paid users promising enhanced intonation and more natural interactions.

Scientists at Berlin’s Charité University Hospital have developed an AI model capable of identifying over 170 tumor types from all human organs. The model, called crossNN, focuses particularly on difficult-to-diagnose brain tumors. It analyzes epigenetic characteristics of tumors, providing highly accurate diagnoses without the need for invasive surgery in some cases. The AI achieves 99.1% accuracy for brain tumors and 97.8% accuracy across all tumor types. Researchers plan to conduct clinical trials across Germany to further test crossNN’s accuracy before potentially introducing it into routine cancer care. 

Nvidia’s newest chips have significantly improved the efficiency of training large artificial intelligence systems. According to data released by MLCommons, the number of chips required to train large language models has decreased dramatically. Nvidia’s Blackwell chips are more than twice as fast as their previous generation Hopper chips on a per-chip basis. The data shows that 2,496 Blackwell chips completed a training test in just 27 minutes, while it took over three times as many previous-generation chips to achieve a faster time.