Claude Sonnet 5 and GPT-5.6 both landed this week, the EU AI Act is 24 days from full enforcement, and a record $510 billion flowed into AI startups in just six months.
Claude Sonnet 5 is live for everyone, and GPT-5.6 is weeks away for the public
Anthropic's Claude Sonnet 5, launched June 30, is now the default model for all free and paid Claude users. It can browse the web, run code, and complete multi-step tasks autonomously, capabilities that previously required larger, more expensive systems. Pricing is $2 per million input tokens through August. Meanwhile, OpenAI's GPT-5.6 is in a limited preview restricted to around 20 vetted organisations, with full public release expected this month. It comes in three tiers (Sol (flagship, focused on coding and science), Terra (balanced), and Luna (fast and affordable)), and offers a massive 1.5 million token context window.
Why it matters: If you use Claude or ChatGPT, you are already on a significantly more capable system this week whether you noticed or not. The gap between what's available free and what required enterprise pricing six months ago has closed significantly.
$510 billion went into AI startups in six months, a record that dwarfs all of last year
Global venture funding hit $510 billion in the first half of 2026, well above the $440 billion invested in all of 2025. Four of the five largest venture rounds ever recorded closed in Q1 2026 alone: OpenAI ($122B), Anthropic ($30B), xAI ($20B), and Waymo ($16B). Nearly 40 AI companies hit unicorn status this year. The boom is now spreading beyond model makers into AI infrastructure, healthcare, defence, and robotics, with non-US companies capturing only about 12% of the total.
Why it matters: When this much capital concentrates in one sector this fast, the tools it funds reach consumers quickly and the job market reshapes around them. The products you'll be using in 2027 are being built right now with this money.
Europe's AI Act goes into full force on August 2: 24 days away
The EU AI Act, which entered law in 2024, reaches full enforcement in just over three weeks. From August 2, any AI-generated content distributed in Europe must carry machine-readable labels, and platforms must proactively disclose synthetic content to users at first interaction. This covers chatbots, AI-generated images, deepfakes, and AI-written articles. Providers and deployers who don't comply face significant fines. Platforms are now in the final sprint to update their systems.
Why it matters: This is the most sweeping AI transparency law in effect anywhere. Since building two separate content pipelines is expensive, most global platforms are likely to apply European standards worldwide, meaning the labeling you see online is about to change wherever you are.
US Senate passes the DEFIANCE Act, deepfake victims can now sue
The DEFIANCE Act (Disrupt Explicit Forged Images and Non-Consensual Edits Act) passed the US Senate unanimously, establishing a federal right of action that lets victims of non-consensual sexually explicit AI deepfakes sue creators, distributors, and platforms that knowingly host the content. It builds directly on the Take It Down Act signed in May 2025 and broadens legal liability up the chain, not just to individual creators. Forty-six US states now have some form of deepfake law.
Why it matters: The window where AI-generated abuse imagery existed in a legal grey area is closing. This shift from vague policy intent to actionable federal law signals that other countries (including India) are watching the enforcement playbook take shape.
Google is training 6 million US teachers in AI, completely free
Google launched the Google AI Educator Series, a free ISTE-aligned training programme open to every US educator. Core sessions are live now, with new content released monthly from September. Microsoft separately published its third annual AI in Education Report showing widespread AI adoption in schools and growing demand for teacher support. Across the US, 71 bills in 27 states are working through legislatures to set rules for AI use in classrooms.
Why it matters: The place where AI skills get normalised at scale is schools. When teachers are trained, students get taught, and the next generation enters the workforce with practical AI literacy as a baseline rather than a differentiator.
India gets a UN AI readiness report and its first domestic AI content rules in the same year
UNESCO and India's Ministry of Electronics and IT (MeitY) released the India AI Readiness Assessment Report, the country's first formal evaluation of its ethical AI landscape, with recommendations on governance, inclusion, and infrastructure. Separately, India's 2026 IT Rules Amendment introduced the concept of Synthetically Generated Information (SGI), the first domestic legal framework that formally defines and governs AI-generated audio, video, and visual content that could be mistaken for real.
Why it matters: India is building its AI ambitions (IndiaAI Mission, GPU capacity, startup ecosystem) and its guardrails (content rules, ethical assessment) at the same time. If you are building or deploying AI products in India, the regulatory landscape is forming right now, the time to engage with it is before it solidifies.
August 2 is a deadline, not a date on a calendar
The EU AI Act's full enforcement is 24 days away. Product teams, platform operators, and anyone distributing content in Europe are now in the last sprint before mandatory compliance. The enforcement mechanism is entering genuinely new territory, nobody knows yet how rigorously regulators will move in the first weeks, or whether this sets a precedent that India, the UK, and the US will follow. The next three weeks will tell us whether Europe's most ambitious tech law has real teeth or becomes another compliance checkbox. Watch how major platforms change what you see online.
Kimi K2.7 arrives in GitHub Copilot + xAI builds voice agents in 2 minutes + NAVER and NVIDIA go sovereign in Korea
Moonshot AI's Kimi K2.7 Code became the first open-weight model available inside GitHub Copilot's model picker, giving developers a free open-source option in a major coding tool. xAI launched its Voice Agent Builder, letting anyone create production-ready phone agents with no code in under two minutes, at $0.05 per minute of audio. And NAVER and NVIDIA announced a major deal to expand sovereign AI infrastructure in Korea to support the next generation of HyperCLOVA X models, a signal that national AI independence is becoming a strategic priority across Asia, not just in India and the EU.