EU enforcement is real now: deepfakes need labels and fines start today: while AI tools got dramatically cheaper, more agentic, and closer to your daily workflow.
EU AI Act enforcement is live: deepfakes must be labeled, chatbots must identify themselves
The European Commission began actively enforcing the EU AI Act on August 2, 2026. The immediate requirements: chatbots must tell users they are talking to an AI, deepfake images and videos must carry a visible label, and AI-generated content must include machine-readable marks. A team of 38 new enforcement staff is already reviewing AI systems. Companies that ignore the rules face fines of up to €15 million or 3% of global annual revenue, whichever is higher.
Why it matters: For three years, the EU AI Act was a law in progress. It is now a law in force. Every app using AI-generated content or conversational AI that reaches EU users must comply, and that includes Indian companies with European customers.
OpenAI cuts GPT-5.6 Luna price by 80%: powerful AI just got dramatically cheaper
On July 30, OpenAI slashed the price of GPT-5.6 Luna, its most cost-efficient flagship model, by 80%. GPT-5.6 Terra also dropped by 20%. The GPT-5.6 family now has three tiers: Sol (the most powerful), Terra (balanced for everyday work), and Luna (the affordable option). Luna now sits at a price point that puts frontier-grade AI within reach of individual developers and small businesses.
Why it matters: A price cut this steep on a flagship model changes what is buildable on a student or small-business budget. If you have been waiting for AI tools to become affordable, this is the moment.
Salesforce rebuilds Slackbot as a full AI agent, your messaging app now takes action
Salesforce launched a completely rebuilt version of Slackbot this week, transforming it from a simple notification system into a fully powered AI agent. The new Slackbot can search enterprise data, draft documents, summarise threads, and take action on behalf of employees, directly inside Slack. Salesforce is positioning Slack as the central nervous system of agentic AI at work.
Why it matters: Hundreds of millions of people use Slack for work. An AI agent embedded where you already spend your day is a bigger shift than a new app you have to download and remember to open.
Microsoft open-sources the "missing layer" of the AI agent stack
Microsoft open-sourced infrastructure tools for managing enterprise AI agent deployments, covering telemetry, memory, traces, and observability. As companies deploy larger fleets of AI agents in production, keeping track of what those agents are doing, where they went wrong, and what they remember has emerged as a critical gap. Microsoft is releasing the tools to fill it for free.
Why it matters: Deploying one AI agent is a weekend project. Deploying a hundred agents safely is a systems engineering challenge. Open-sourcing this layer lowers the barrier for companies building real agent-powered products.
Google's voice agents can now call stores, check inventory, and complete purchases for you
Google is rolling out consumer AI agents that autonomously make phone calls on the user's behalf, calling a store to check stock, asking about hours, or completing a purchase, all without the user picking up the phone. This extends Google's agentic capabilities from digital tasks into real-world phone commerce.
Why it matters: When an AI can pick up the phone for you, the question of how businesses receive and respond to AI-generated calls becomes urgent. This is the consumer face of the agentic AI shift, and it is arriving faster than most people expected.
Maharashtra AI Policy 2026: Rs 10,000 crore investment, 2 lakh people trained by 2031
Maharashtra announced its AI Policy 2026, setting a target of attracting Rs 10,000 crore in investment and creating over 1.5 lakh jobs by 2031. The policy establishes a Maharashtra AI Mission, proposes AI Cities in five innovation zones, and includes a 20% subsidy for 5,000 MSMEs adopting AI. A central platform called the Maha AI Tools Hub will connect businesses to approved AI tools. Two lakh students and professionals are targeted for AI training.
Why it matters: State-level AI policy with actual numbers and a subsidy scheme for small businesses is rare. If you run or work for an MSME in Maharashtra, there is now a government programme specifically designed to pay for part of your AI adoption cost.
The EU enforcement era is the new normal, and the ripple effects are global
EU AI Act enforcement is not just a European story. Any product that reaches EU users (including apps built in India, the US, or anywhere else) must comply with the new transparency rules. The enforcement team is actively reviewing AI systems. Watch for the first major fine: it will set the standard for what "compliance" actually means in practice, and every AI team in the world will be reading that decision.
Alibaba's 2.4-trillion-parameter model + AI startup funding hits $337B in 2026
Alibaba unveiled Qwen3.8-Max, a 2.4-trillion-parameter model built for long-horizon coding and agentic workflows, with open weights planned. Separately, AI startup funding in 2026 has already reached $337 billion across nearly 400 deals: Anthropic closed a $65B round at a $965B valuation, briefly making it the most valuable private company on the planet.