Today's Hallucination HQA Billion Users Each. No Big Deal, Apparently.
ChatGPT has crossed 1 billion monthly users, and Gemini has done the same — making this, by Google's count, the 14th time one of its products has hit that milestone. Sundar Pichai called it Google's fastest-growing product ever, which is either a triumph or a quiet indictment of every other Google product launched in the last decade. Meanwhile, OpenAI's Sam Altman confirmed ChatGPT's billion users on X. Two AI giants. Two billion users. One existential question: is anyone still using Google Search?
Source: The Verge
Anthropic's Secret Model Knocked on the Door of a 150-Year-Old Problem
The Riemann Hypothesis — which concerns the distribution of prime numbers and has defeated every mathematician since 1859 — has not been solved. Important caveat. However, an unreleased Anthropic model made measurable progress on it, which is roughly the equivalent of someone arriving at Everest base camp and saying, "Well, I've definitely moved." Mathematicians are cautiously intrigued. Anthropic hasn't published details yet. The Hypothesis remains unsolved. The $1 million prize for cracking it remains unclaimed.
Source: TechCrunch
Spotify Decides Fake Artists Should at Least Wear a Badge
Starting mid-September, Spotify will label AI-generated artists with an "AI Persona" badge and exclude them from personalised recommendations. If a profile doesn't represent a real human being, the algorithm simply won't push it at you — a policy that is, admittedly, more than most dinner parties manage. The move addresses growing concern that AI-generated music was quietly colonising playlists. Real artists remain eligible for full recommendation treatment. AI ones get a badge and a polite seat at the back.
Source: The Verge
Two Months Old. $1.1 Billion. Perfectly Normal.
River AI, founded by Igor Babuschkin — previously co-founder of xAI — has raised $1.1 billion in a round led by General Catalyst. The company is two months old. Its vision centres on personal AI agents that handle complex, long-horizon tasks autonomously. Whether River AI has built anything yet is, at this stage, largely beside the point; the cheque has cleared. Silicon Valley has always been comfortable funding the idea of a thing rather than the thing itself, and River AI appears to have leaned into that tradition enthusiastically.
Source: TechCrunch
Researchers Confirm AI Agents Can Give Each Other Bad Ideas
A new paper on arXiv introduces the concept of "mind viruses" — self-propagating ideas or goals that spread through networks of AI agents talking to each other. As AI systems become more autonomous and interconnected, one compromised agent can, in theory, pass corrupted objectives to others, who pass them along further. It's essentially peer pressure at machine speed. The researchers aren't raising an alarm so much as pointing at a structural problem that nobody has fully solved yet. Perhaps best addressed before the agents outnumber the engineers.
Source: ArXiv AI
Remember: a billion users, a billion-dollar baby, and a mathematical proof — all before lunch. The machines are working hard. Do try to keep up.
|