Baidu is strengthening its Foundation Model Unit (BMU) with two key hires. Sun Tianxiang, BMU head, brought in a researcher codenamed Wuqin from a North American Frontier Lab to boost ERNIE pre-training. Meanwhile, former DeepSeek researcher Wei Haoran, lead author of DeepSeek-OCR, joined BMU as head of multimodal algorithms. Wei previously contributed to the open-source Unlimited OCR, which set a record on OmniDocBench. The moves come as Baidu ramps up core model development.

Prediction markets are expanding at a pace that is hard to ignore, with sports contracts driving much of the recent surge. The article argues that these venues may offer genuine value beyond speculation: more direct hedging tools, cleaner probability signals, and new ways to price outcomes that traditional financial markets or insurers often cannot handle. It points to examples tied to Federal Reserve rate decisions, corporate hedging, election outcomes, entertainment, and even future GPU rental prices. At the same time, the growth story is colliding with a widening regulatory and consumer-protection debate in the U.S. Kalshi and Polymarket are facing legal action from states, tribes, and private parties, while the Commodity Futures Trading Commission is asserting federal authority over event contracts. The conflict is not only about jurisdiction. It is also about whether sports event contracts function, in practice, like sports betting products that compete for the same users while operating under a different rule set. The piece also highlights uneven outcomes for retail users, concentration of profits among sophisticated traders, the rise of parlay-style products, questions around insider trading, and the lack of consistent safeguards for younger or vulnerable participants. Its central claim is that prediction markets may indeed be better markets in some respects, but they have not yet become better protected ones.

AI model names often reveal more than branding. From Claude honoring Claude Shannon to Grok borrowing a sci-fi term and Doubao using the plain name of a snack, naming choices reflect each team's ideas about technology and product identity.

Kimi (Moonshot AI) secured $2 billion in funding, pushing its valuation above $20 billion, led by Meituan Dragonfly. This marks the fourth round in six months, with total capital raised exceeding $3.9 billion and valuation quadrupling since November. Cumulative funding surpasses 37.6 billion yuan, a record for AI startup.