US midterms are one week away and AI deepfakes are already everywhere, India's AI regulation paper lands within days, and consumer AI funding hit a new high.
One week to US midterms — AI deepfakes are already everywhere
The Brennan Center retested six major AI image generators and found that ChatGPT still converted nearly all election-disinformation prompts into political imagery. With US midterm elections seven days away, researchers and platforms are watching for AI-generated ads, synthetic candidate speeches, and fake polling images. Thirty-one US states have now enacted laws against political deepfakes, but enforcement is largely untested at this scale.
Why it matters: This is the first major election cycle after the TAKE IT DOWN Act passed. How well the rules hold under real pressure will shape AI legislation everywhere — including India's own upcoming paper.
India's AI regulation paper: T-minus days
IT Minister Ashwini Vaishnaw said on October 8 that MeitY's AI regulation consultation paper would be published within a month — putting the deadline at November 7. Priorities named so far are safety, deepfakes, user harm, and a human-first approach. Separately, an empanelment route for government AI procurement will let Indian startups bid on departmental projects without going through single-tender processes.
Why it matters: India's 1.4 billion people have had no formal AI rules until now. Whatever the paper says will define the guardrails for every product, employer, and student who uses AI in India.
Wabi raises $55M to build AI for things people do together
Wabi raised $55 million at a $1.25 billion valuation to develop AI assistants for shared everyday activities — splitting bills, planning trips, coordinating household tasks. Investors include a16z, IVP, Forerunner, and the Amazon Alexa Fund. The round reflects a clear bet that consumer AI will move beyond solo productivity tools into socially embedded ones.
Why it matters: Most AI tools are built for one person sitting alone at a laptop. Wabi is building for the way people actually live — together. If it works, it changes what an AI assistant looks and feels like.
Manus raises $500M to build AI that builds things for you
Manus, which builds AI agents that can make full apps, websites, and presentations on demand, raised $500 million led by Boyu Capital. The company sits at the intersection of two trends: the shift from AI as an assistant to AI as a builder, and growing demand from small teams who cannot hire developers. The round values the company in the multi-billion range.
Why it matters: If you have an idea for a product or a website and no technical background, tools like Manus are shortening the path from idea to live thing dramatically. This funding means the category is being taken very seriously.
Nirva raises $25M for an AI pendant that listens only to you
Nirva's wearable AI pendant raised $25 million in late October. The device listens only to its wearer — not nearby conversations — and surfaces relevant information and reminders through a companion app. No screen, no summons. It joins a small but growing category of ambient AI hardware that wants to sit in the background of daily life rather than demand attention.
Why it matters: The form factor for AI is still being figured out. Nirva's approach — audio-only, always passive, private by design — is one version of what ambient AI looks like when it is not trying to be a phone.
NVIDIA releases open models for enterprise agents and local PCs
NVIDIA released a new set of open models designed to run AI agents in enterprise environments and on consumer PCs — no cloud dependency required. A separate open transcription model was also released. Both are available for developers to build on. The timing follows NVIDIA's $12.9 billion Hugging Face acquisition, which is still under DOJ review.
Why it matters: Open models that run locally mean private data stays private. For businesses in regulated industries, and for users in countries with data sovereignty concerns, local-first AI is not a niche preference — it is a requirement.
The midterm deepfake test — and what it means for India
Seven days from now, the US midterm elections will be the first major democratic vote to happen after the TAKE IT DOWN Act, after 31 states passed anti-deepfake laws, and after every major platform updated its AI content policies. Whether AI-generated political content actually distorts the outcome — or is contained — will either accelerate or slow down similar legislation globally. India's own AI regulation paper lands in the same window. The two events are on a collision course with the same question: does regulation keep pace with the tool?
Cohere closes a new round + Claude Duet research preview goes live
Cohere, which sells enterprise AI tools for private deployments, closed a new funding round on October 30 (final amount unconfirmed). Separately, Anthropic opened a research preview of Claude Duet — an autonomous agent designed for longer tasks — on iOS, Android, and web. Neither story is a blockbuster on its own, but both signal that enterprise and agentic AI are both attracting capital and shipping product at the same time.