AI News from November 9, 2026

Nominal raised $240 million at a $3 billion valuation, a sharp re-rating for its hardware-testing business as AI infrastructure spending accelerates.

Cisco says AI Defense will add Anthropic’s Claude Mythos for AI red teaming, linking a security platform with model-led testing for safer deployments.

Meta’s Muse Ember targets lower-cost agent execution with multimodal inputs, giving developers a cheaper model for repetitive tool calls, coding, and routing.

Tencent AI chief Vinces Yao warns that AI-enabled bioweapons are approaching and says safer open models require training-data controls, not just refusals.

Reports say Apple and Samsung are looking to Chinese AI firms, a move that could reshape smartphone AI partnerships amid intensifying competition.

Israeli start-up Zeroport is pursuing physical safeguards for AI agents, raising questions about how hardware could constrain autonomous software in business.

NVIDIA is investing in Korean AI startup Point2 and plans three more deals, signaling deeper ties with Seoul’s AI ecosystem despite limited detail.

NVIDIA plans investments in four Korean AI startups, extending its Korea strategy across AI chips, models, robotics, enterprise software and infrastructure.

PhocusWire has named Blis AI among its Hot 25 Travel Startups for 2027, putting the company on a watchlist for travel-focused AI buyers and builders.

A report says Korean firms are evaluating NVIDIA tools for secure AI agents, but limited evidence leaves the products, partners, and plans unconfirmed.

LG AI Research is opening free API access to K-EXAONE 2.5, giving developers a new way to test South Korea’s national AI model in practice.

Anthropic’s reported Dreaming system lets AI agents reorganize memory offline, with XenoSpectrum citing a rise in complex research scores from 73% to 87%.