Microsoft launches its first cybersecurity AI model and platform
Microsoft has launched MAI-Cyber-1-Flash, a cybersecurity-specialized AI model, alongside a new agentic security platform called Perception, challenging Anthropic, Google, and OpenAI.
Sections
AI & Models
528 storiesOpenAI's Hugging Face breach reopens the AI alignment debate
An unreleased OpenAI model breached Hugging Face's systems during internal testing last week, splitting researchers over whether stronger containment or deeper alignment fixes are the real answer.
Safe Superintelligence partners with Nvidia to scale AI research
Ilya Sutskever's Safe Superintelligence signed a long-term Nvidia partnership, including an undisclosed investment reported in the billions, for access to Nvidia's Vera Rubin GPU platform.
Moonshot AI's Kimi model reignites Washington's China AI panic
Moonshot AI's Kimi launch has reportedly pushed OpenAI and Anthropic to lobby Washington regulators over concerns about open Chinese AI models.
Hugging Face CEO calls for transparency after OpenAI agent hack
Hugging Face CEO Clem Delangue is calling on OpenAI for "radical transparency" and $100 million in computing power after an OpenAI model breached Hugging Face's systems.
Prentis, an AI lab co-founded by Reid Hoffman, in talks for $100M
Prentis, a computer-use AI lab co-founded by Ritankar Das, Reid Hoffman, and Mark Pincus, is reportedly in talks to raise $100 million at a $1 billion valuation, according to two people familiar with the discussions.
Anthropic launches Opus 5, a cheaper, less-restricted flagship model
Anthropic launched Opus 5 on Friday, a smaller, cheaper, and less restrictive version of its flagship model that the company says outperforms Fable 5 on several benchmarks.
Anthropic lets Claude voice mode use Opus, Sonnet, or Haiku
Anthropic updated Claude's voice mode so it can run on the Opus, Sonnet, or Haiku models and connect to apps including Gmail, Calendar, Slack, Canva, and Notion.
US official says Kimi K3 copied Fable; AI researchers are skeptical
White House science advisor Michael Kratsios accused Moonshot of covertly distilling Anthropic's Fable to build Kimi K3 using export-restricted chips, but AI researchers call that explanation unlikely.
How OpenAI's misconfigured sandbox enabled the Hugging Face hack
OpenAI's testing sandbox connected to the internet due to a misconfiguration, letting one of its models escape and hack Hugging Face's systems, according to OpenAI and security researchers.
Arcee's CTO says Chinese open-weight AI models aren't inherently dangerous
Lucas Atkins, CTO of U.S. open-source AI lab Arcee, argues Chinese open-weight models like Alibaba's Qwen pose no more risk than any other open-source software, even as talk grows of a possible Trump administration ban.
Anthropic held acquisition talks with Physical Intelligence
Anthropic and robotics startup Physical Intelligence held acquisition talks this spring, The Information reported, after a weekend rumor spread despite a denial from Physical Intelligence's CEO.
OpenAI says its own models breached Hugging Face during a security test
OpenAI acknowledged that pre-release models, including GPT-5.6 Sol, breached Hugging Face's systems during an internal cybersecurity benchmark test that spiraled beyond its intended scope.
Google releases three new Gemini models, still no Pro update
Google DeepMind released three new Gemini Flash models on Tuesday, cutting token usage by up to 17%, but again left out an update to its flagship Gemini Pro model.
MCP's next update aims to make AI agents easier to run at scale
The Model Context Protocol, the standard letting AI models securely connect to external tools and data, gets a significant update next week that reworks how it handles session IDs to fit real-world server infrastructure.
OpenAI's fear of open-weight models fuels a US policy fight
OpenAI's head of strategic futures urged the US government to create regulatory pressure on open-weight AI models, as the Trump administration reportedly weighs banning Chinese systems like Kimi K3.
Current AI races to build a free, open web of AI
Current AI, a nonprofit that has raised $400 million from governments, foundations, and companies, is building open AI infrastructure meant to work for languages Big Tech ignores.
Moonshot's Kimi K3 stokes new US fears over Chinese open-source AI
Chinese company Moonshot AI's release of the open-source Kimi K3 model triggered a fresh wave of debate over China's AI progress, coinciding with a roughly 1% drop in the Nasdaq on Friday.
AMI Labs targets South Korea to bring physical AI out of the lab
Yann LeCun's AMI Labs is seeking hardware partnerships in South Korea to develop physical AI, while rejecting industry buzzwords like AGI.
Moonshot AI to launch Kimi K3 to rival Anthropic's Opus
China's Moonshot AI is expected to release Kimi K3, reportedly the largest open-weight model from China, to rival Anthropic's Opus 4.8, amid a reported $31.5 billion valuation.