Fiduciary AI agents: Vijil says trust must be tested in production
Vijil CEO Vin Sharma argues enterprise AI agents need continuous trust scoring as benchmarks miss production failures and collusio...
By Wei-Lin Zhao 1 week agoSection
AI
Nvidia SSI investment moves Sutskever’s lab toward Nvidia chips
Nvidia is investing an undisclosed “substantial” sum in SSI, giving Ilya Sutskever’s AI lab access to its Vera Rubin GPU platform.
By Colin Brandt 1 week agoSnowflake Cortex AI Gateway aims to govern enterprise AI agents
Snowflake is adding a control layer for AI agents, with identity partners helping limit data access, audit actions and track costs...
By Colin Brandt 1 week ago
Nvidia employee detained in Taiwan chip smuggling probe, reports say
Taiwan prosecutors detained a reported Nvidia employee as they widen a probe into alleged illegal exports of AI servers to China.
By Renata Fuchs 1 week ago
Amodei open weights stance rejects ban while warning on AI risks
Anthropic CEO Dario Amodei says open-weight AI should not be banned, but argues powerful releases need safety tests and export con...
By Colin Brandt 1 week agoModel Context Protocol update makes MCP stateless for enterprise agents
The AAIF release changes MCP architecture, auth and extensions as backers push AI agents toward large production deployments.
By Colin Brandt 1 week agoKimi K3 license complicates Moonshot AI’s open-weights release
Moonshot AI released Kimi K3 weights, but commercial users face revenue, service and attribution conditions under its custom licen...
By Colin Brandt 1 week ago
Kimi K3 open weights released by Moonshot AI with infra code
Moonshot AI released Kimi K3 weights and selected infrastructure, giving developers a closer look at a Chinese model near frontier...
By Renata Fuchs 1 week ago
ChatGPT task crossover rises in OpenAI analysis of work use
OpenAI says 43.5% of job-specific ChatGPT work prompts it studied involved tasks tied to another profession.
By Renata Fuchs 1 week agoMicrosoft AI cybersecurity model targets enterprise security costs
Microsoft says MAI-Cyber-1-Flash and Project Perception can cut MDASH costs about 50% by routing harder security tasks to GPT-5.4.
By Colin Brandt 1 week ago
Microsoft MAI-Cyber-1-Flash launches for AI security workflows
Microsoft says its new cybersecurity model helps MDASH cut AI security costs by routing only harder work to GPT-5.4.
By Colin Brandt 1 week ago
OpenAI wins early ruling in ANI copyright case in Delhi
Delhi High Court refused ANI’s bid to block OpenAI, finding no proof ChatGPT copied articles verbatim at this stage.
By Wei-Lin Zhao 1 week ago