Anthropic's Claude Managed Agents can now "dream," sort of
Also, 5-hour usage limits will double for Pro and Max users of Claude Code.

Also, 5-hour usage limits will double for Pro and Max users of Claude Code.

French AI startup ZML has launched ZML/LLMD, an inference server designed to optimize large language models across various chips, including Nvidia, AMD, and Apple. The software aims to break vendor lock-in and reduce AI costs by enabling efficient use of diverse hardware, potentially disrupting Nvidia's market dominance.

Researchers have devised HalluSquatting, a novel pull-based AI attack that exploits LLM hallucinations of resource identifiers to create massive botnets, conduct large-scale DDoS attacks, and infect devices. The attack targets AI coding assistants and agents by registering predicted hallucinated identifiers with malicious code.

AI chip company SambaNova Systems has secured $1 billion in its Series F funding round, valuing the company at $11 billion. Led by General Atlantic, the round saw participation from new and existing investors, including Intel and JPMorgan Chase, which selected SambaNova as an inference-infrastructure partner.

Meta launched Muse Image, an AI image generator available on Meta AI, Instagram Stories, and WhatsApp. A controversial feature allows users to edit public Instagram profiles' images without explicit consent, raising privacy alarms.

Meta has launched Muse Image, a free AI image generator from its Superintelligence Labs, available via the Meta AI app, Instagram Stories, and WhatsApp. It offers preset prompts, custom ad creation, interior design visualization for Facebook Marketplace, and prompt-based editing.

Decagon CEO Jesse Zhang proposes open source and frontier AI models are not rivals but part of a lifecycle. Mature use cases shift to cheaper open source, while expensive frontier models drive new discoveries, maintaining overall spend.