Daily AI News | Google launches Gemini Omni 1.1 Flash for controllable video generation and editing
# Daily AI News | Google launches Gemini Omni 1.1 Flash for controllable video generation and editing
## Lead Story
### Google launches Gemini Omni 1.1 Flash for controllable video generation and editing
Google DeepMind launched Gemini Omni 1.1 Flash with scene extension up to 40 seconds, start-and-end-frame interpolation, 360p previews, and 4K upscaling. It is available through the Gemini API, Google AI Studio, and Flow.
**AI take:** The important shift is workflow control, not resolution alone; long-shot consistency and API economics will determine production value.
## More News
### More than 100 companies call for coordinated defenses against rogue AI
More than 100 AI, cloud, and security companies issued an open call for companies and governments to coordinate defenses, threat sharing, and responses to malicious AI-enabled cyber activity.
**AI take:** The test is whether the coalition turns broad agreement into shared standards, incident reporting, and auditable controls.
### Anthropic previews a Model Hardware Standard for physical agents
Anthropic released a research preview of the Model Hardware Standard, a shared specification and driver layer for laboratory and manufacturing devices with MCP, command-line, and API access. Initial partners will test compatibility, with open sourcing planned later.
**AI take:** Physical agents are constrained by fragmented device protocols and permissions as much as reasoning; ecosystem adoption will matter more than a single demo.
### Google pilots double-blind AI evaluations for private models and benchmarks
Google DeepMind used Confidential Space to pilot evaluations where the model provider cannot see private prompts and the evaluator cannot see model weights. Singapore AISI, OpenMined, AVERI, and MLCommons participated.
**AI take:** The approach could reduce both benchmark leakage and model-IP exposure, giving regulators and independent evaluators a more credible testing path.
### ChatGPT ads are coming to Free and Go users in India
OpenAI said it will show ads to ChatGPT Free and Go users in India, beginning with about 50 brands and followed by an ad manager. Higher-paid tiers are outside this rollout.
**AI take:** India is becoming a major monetization test; separation from answers, privacy boundaries, and user trust will matter most.
### AI startup Instinct raises a $250 million Series B
TechCrunch reported that Instinct raised a $250 million Series B, bringing total funding to $350 million at a roughly $2.5 billion valuation. Index Ventures and Benchmark co-led the round while the product remains in private beta.
**AI take:** The financing shows continued appetite for new consumer-AI interfaces, but retention and revenue must eventually support the valuation.
### Lambda closes a $926 million loan facility for contracted GPU deployment
AI cloud provider Lambda closed a $926 million senior secured Term Loan B to deploy GPUs for a committed investment-grade customer. The facility carries a Moody’s Baa2 rating and matures at the end of 2030.
**AI take:** Long-term customer contracts are increasingly financing compute buildouts, while concentrating execution and hardware-cycle risk.
### LTX-2.5 ships with open weights
LTX announced through its official account that LTX-2.5 now ships with open weights, allowing developers and creators to inspect, adapt, and deploy the model beyond the hosted product.
**AI take:** Open weights broaden deployment and customization; licensing, hardware requirements, and community tooling will determine adoption.
### Salesforce introduces Slack Code for team-based coding agents
Salesforce introduced Slack Code, using code channels to put developers and coding agents in one Slack workspace. It named Anthropic, Cognition, GitHub, ChatGPT, and Vercel among partners, with APIs planned later.
**AI take:** Coding agents are moving from individual IDEs into team coordination; permissions, review trails, and deployment context will determine real adoption.
### SK hynix breaks ground on a $4 billion-plus Indiana HBM facility
SK hynix broke ground on an Indiana HBM production base costing more than $4 billion. It targets mass production of U.S.-made HBM in the second half of 2029 and about 7,000 jobs.
**AI take:** The HBM supply chain is extending into U.S. production and packaging, but a 2029 ramp will not resolve near-term shortages.
## Rumors
### NVIDIA reportedly nears a $12.9 billion agreement to acquire Hugging Face
TechCrunch and The Information reported that NVIDIA may be close to, or have agreed on, a roughly $12.9 billion acquisition of Hugging Face. Other reporting cautioned that documents were not signed and talks could still fail; neither company had confirmed the deal.
**AI take:** A completed deal could join a central open-model platform with the dominant AI-chip supplier, but it remains attributed reporting until primary confirmation.
