Golem · GLM / Jadawalin farashi

Prime Intellect reasoning platform is online: nearly 1 trillion Tokens have been processed internally every day

Anthropic upgrades Claude's anti-distillation defenses, reducing model inference data leakage by 90% to 99%

Anthropic: Small AI models may gain the ability to self-replicate via smart devices within months.

L2 Cools Down, PoW+AI Rises? Analyzing TOKEN2049 Spotlight Project Pearl

PRL public chain replaces traditional hash computation with AI computing power mining. As L2s shut down one after another, can the PoW narrative be revived thro...

When Agents Learn to "Collude": As AI Gets Smarter, How Do We Draw the Safety Boundaries?

In the AI era, new security issues are shifting from "preventing a single Agent from exceeding its authority" to "how to prevent a group of Agents from collecti...

The kill line for Chinese large models has been killed.

Anthropic Haiku 5.5 cuts inference costs to one-tenth of the previous generation. The price war among U.S. AI giants has battered Chinese large-model concept st...

PA Daily | Binance Launches AI Product Stack Binance Intelligence; CFTC Proposes First Round of Crypto Market Rules

Important information last night and this morning (October 4th - October 5th)

Xie Jiayin: Bitget’s reserve ratio for 19 mainstream assets reached 131%, and the user protection fund increased to US$316 million. Bitget’s latest reserve cert...