join.dev
AI news · 10 stories
Anthropic releases Claude Haiku 5.5 with adjustable effort · Microsoft brings hybrid AI and agent controls to Windows · Anthropic launches Cyber Mission for infrastructure and open source
Anthropic released Haiku 5.5 for high-volume, cost-sensitive work, adding adjustable effort controls to its small-model line. It is available across major cloud platforms, giving developers another option for summaries, classification and narrowly scoped agent tasks.
Read at Anthropic
Microsoft announced Windows capabilities that combine local and cloud AI, alongside generally available Microsoft Execution Containers for agent isolation. Planned Copilot access to local context requires permission, while GitHub tools are due to receive an experimental hybrid-intelligence preview later in October.
Read at Microsoft
Anthropic launched its Cyber Mission with a Critical Infrastructure Defense Program combining Claude models, on-site engineers and threat research. It also introduced free, opt-in OSS Scanner checks for open-source projects, whose model-generated findings are delivered without human review.
Read at Anthropic
Anthropic committed $150 million over three years to the federal Genesis Mission for AI-assisted scientific discovery. The commitment will make Claude available to more than 15 participating agencies, including NASA, NIH and NSF, expanding access across public research.
Read at Anthropic
Meta described new language-model detection for seemingly benign ads that direct users toward child exploitation material elsewhere. It also added destination analysis and an AI red-teaming agent to probe its defenses, addressing evasion tactics beyond the visible ad itself.
Read at Meta
Liquid AI released d1-3B and experimental d1-omni-600M as open-weight decision models for edge devices. They support text with images, or text with audio in the smaller model, and return decisions in one forward pass rather than generating token sequences.
Read at Liquid AI
Nous Research announced a $90 million Series B led by Robot Ventures, with Nvidia and Samsung among the participants, valuing the company at $1.5 billion. The financing backs the developer of Hermes, an open-source agent that can interact with users’ files.
Read at SiliconANGLE
Manus parent Butterfly Effect announced a $500 million funding round led by Boyu Capital and IDG Capital after China blocked Meta’s acquisition. The company says it is hiring domestically and internationally, supporting continued development as an independent AI-agent business.
Read at PYMNTS
Mecka announced a $60 million Series B led by Sequoia Capital, with new investors including Nvidia, M12, Qualcomm Ventures and Samsung. The funding will expand data infrastructure, internal research and commercial robot deployment, addressing the need for real-world training data.
Read at Mecka
Anthropic published a Usage Policy update taking effect November 12, consolidating restrictions on deceptive campaigns and clarifying high-risk uses. It adds requirements for potentially injurious autonomous hardware, including qualified operator oversight, a stop capability and a safe state when Claude disconnects.
Read at Anthropic