Google's robotics and DNA models, Anthropic's new workplace tools, and New York's AI safety law taking effect — a full day across labs, policy, and products.
Google DeepMind drops Gemini Robotics 2.5 and a new DNA prediction model in one go
DeepMind released Gemini Robotics 2.5, its most capable robotics AI yet, now rolling out through partners including Boston Dynamics and Apptronik. The same announcement introduced AlphaGenome 2, a model that predicts how DNA sequences affect gene activity — not approved for clinical use, but described as a research tool for drug and disease scientists.
Why it matters: Two completely different fields in one announcement. AI is not just moving fast inside software — it is reaching into biology and physical machines at the same time. If you work in biotech, pharma, or any physical industry, this is the direction the technology is heading.
Anthropic rolls out Claude Sheets, Meetings, and Whiteboards — AI built into how you work
Following its San Francisco event earlier this week, Anthropic is opening Claude Sheets (beta, all paid plans, exports to Excel), Claude Meetings, and Claude Whiteboards to users. Claude Fable 5.2 also shipped, positioned as Anthropic's new top-end model focused on coding, agents, and complex knowledge work.
Why it matters: These tools bring AI into documents, meetings, and collaborative work instead of keeping it in a chat window. The gap between people who know how to use AI in their daily workflow and those who don't is about to get more visible.
New York's RAISE Act takes effect — safety rules and incident reporting now required for AI builders
New York's RAISE Act, modeled partly on California's AI rules, is now in force. It requires companies developing AI systems to follow documented safety protocols and report serious incidents. It joins a growing patchwork of state-level rules as the US federal government has passed only one AI-specific law to date.
Why it matters: State laws are where AI regulation is actually happening in the US right now. If you build, deploy, or manage AI tools for work, it is worth knowing whether the RAISE Act applies to your team or vendor.
EU deepfake prohibition is weeks away — platforms scrambling to comply
The EU's AI Act deepfake labeling requirements have been in effect since August 2026. On December 2, the next phase kicks in: a prohibition on AI-generated non-consensual sexual deepfakes and AI-generated child sexual abuse material. Thirty-three US states also now have some form of deepfake protection on the books, though courts have struck down two state election-deepfake laws on First Amendment grounds.
Why it matters: Enforcement is no longer theoretical. Platforms, creators, and AI developers in the EU are on a countdown. The US picture is messier, but direction is clear: governments are moving from discussing AI harms to acting on them.
Indian AI startups raised $676M in H1 2026 — four times more than the year before
Inc42's latest count puts Indian AI startup funding at $676 million in the first half of 2026, with deal count up 90% year on year. Recent closes include Simplismart's $14M Series B and Venomics AI, an IISc spinout working on drug discovery, raising $2.7M. India's overall startup funding fell 9% in the same period — AI is the exception pulling the numbers up.
Why it matters: India's AI ecosystem is growing fast relative to the broader market. Deals are still closing, the government is pushing hard on infrastructure, and the window to build in this space is open. The gap with the US remains large — but that gap is exactly where the opportunity is.
Astral raises $35M to personalise learning with AI — biggest ed-tech AI round of the month
Astral, an AI-driven education platform, closed a $35M round on November 9, focused on hyper-personalised learning experiences for students and lifelong learners. It is the largest AI education funding round tracked this month, following earlier 2026 rounds from Gizmo ($22M), SchoolAI, and others in the AI tutoring and adaptive learning space.
Why it matters: Education is now one of the fastest-growing segments in AI startup investment. Tools designed around how you actually learn — not just how you search or chat — are getting serious money. If you are a student or educator, this is the category worth watching.
December 2 is the EU's AI enforcement moment — and it sets the global tone
Three weeks from now, the EU moves from labeling requirements to outright prohibition on AI-generated non-consensual intimate imagery and AI-generated CSAM. It is the first major hard ban under the AI Act, and every other government watching the EU's approach will be reading the enforcement closely. Whether platforms comply cleanly, challenge in court, or quietly scramble will tell us a lot about how the next wave of AI regulation lands globally.
Meta Muse Ember (smallest efficient model for tool calls) + OpenAI launches Businesses in ChatGPT
Meta released Muse Ember, its most lightweight Muse model, designed specifically for tool calls, coding subtasks, routing, and extraction — a signal that efficiency at the small end of the model spectrum matters as much as raw capability. Separately, OpenAI opened Businesses in ChatGPT to beta users, adding workspace and team features to the consumer product.