Google I/O Day 2 delivered a cascade of launches: Gemini 3.5 Flash in Search for everyone, AI glasses arriving this fall, and a $100/month price cut that could change who gets access to frontier AI.
Gemini 3.5 Flash is live: faster than a flagship at half the cost
Google officially launched Gemini 3.5 Flash at I/O 2026, calling it the first model that truly combines frontier-level intelligence with real-world action. It outperforms Gemini 3.1 Pro on tough coding and agentic benchmarks, and it's already the default model powering AI Mode in Google Search for everyone globally. It's also available to developers through Google AI Studio and Android Studio today.
Why it matters: When the smartest search engine on earth upgrades its brain, billions of people feel it immediately, even if they don't know a model was swapped. Tasks that used to take a developer days can now complete in hours, at a fraction of the cost.
Google's AI glasses are real, and coming to your face this fall
Google revealed Intelligent Eyewear at I/O 2026: two types of smart glasses, audio and display. Audio glasses (launching first, this fall) give you Gemini in your ear for live navigation, real-time voice translation, and answering questions about what you're looking at. They're made in partnership with Samsung, Warby Parker, and Gentle Monster, and work with both Android and iOS. Display glasses are in development for later.
Why it matters: This is the form factor that makes AI truly ambient, present without needing to be summoned. The combination of fashion-brand partnerships and cross-platform compatibility suggests Google is serious about mainstream adoption, not just a tech demo.
Google AI Ultra drops from $250 to $100 a month, and adds a 24/7 AI agent
Google slashed the price of its AI Ultra subscription from $250 to $100 per month and reworked the limits: instead of a daily cap on prompts, you now get a compute budget that refreshes every five hours. The tier also includes Gemini Spark, a new personal AI agent that monitors your Gmail, reviews your bank statements for hidden subscriptions, and synthesises your documents, all inside a private, Google-account-aware context.
Why it matters: A 60% price cut on the most capable AI subscription tier is a significant access shift. It's still not cheap, but the gap between power users and everyone else just got a lot smaller.
SynthID goes industry-wide: OpenAI, ElevenLabs, and Kakao now watermark their AI content
Google confirmed at I/O that OpenAI, ElevenLabs, and Kakao have all adopted SynthID, its invisible watermarking technology for AI-generated content. The EU AI Act's Article 50 transparency requirements take effect in August 2026, and watermarking obligations for content providers were just pushed back to December 2026 following a May agreement. The industry is converging on a two-layer standard: C2PA metadata credentials plus imperceptible watermarks like SynthID.
Why it matters: When OpenAI and Google agree on a content-labelling standard, it becomes the de facto global default. This is the foundation of being able to tell real from AI-generated, a problem that affects elections, hiring, journalism, and education.
1 in 3 job candidates may already be a deepfake. HR is not ready.
A Pindrop Security analysis of 300 job applicant profiles found over a third were entirely fabricated: AI-generated resumes, deepfake video interviews, the works. Gartner projects that by 2028, one in four candidate profiles globally will be fake. The US DEFIANCE Act, passed in January 2026, creates a federal right to sue creators and distributors of non-consensual deepfakes, but that covers personal harm, not employment fraud. No federal law specifically addresses fake job applicants yet.
Why it matters: This is already happening at scale. If you manage hiring, train HR teams, or are looking for work yourself, the rules of the game have changed and most organisations are still playing by the old ones.
India's AI skilling machine switches on: millions of workers in training
India's AI workforce push is moving from announcements to action. Microsoft has committed to 2 million AI skilling opportunities in India. TCS is training 350,000 employees; Wipro, 220,000. The government has established five National Centres of Excellence for Skilling, and FutureSkills PRIME has 1.85 million registrants. All of this sits inside a Rs 10,371 crore IndiaAI Mission budget and a macro bet that AI adoption could lift India's GDP growth rate from 5.7% to 8%.
Why it matters: Scale is the word. The numbers are large enough that if even half of these commitments deliver, India produces the biggest AI-capable workforce in the world. For students and early-career professionals, the opportunity window is open right now.
Who actually benefits when Google makes AI cheaper?
Google cutting AI Ultra from $250 to $100 is a big deal, but $100 a month is still more than many people in India, Southeast Asia, and across the Global South spend on all software combined. The capability gap between the free tier and the paid tier is growing, not shrinking: Gemini Spark, the personal agent, is Ultra-only. Watch whether the price floor continues to fall, whether free-tier capabilities improve meaningfully, and whether India-specific or student-specific pricing emerges. The access story matters as much as the technology story.
EU pushes watermarking deadline + AI agents raise $230M + Gemini 3.5 Pro is next
The EU shifted its AI content watermarking obligation from August to December 2026 after a May agreement, giving platforms more time to comply. Parallel, which builds web-search infrastructure for AI agents, raised $230M at a $2B valuation, a sign that the plumbing behind autonomous agents is attracting serious capital. And Google confirmed Gemini 3.5 Pro is already in internal testing, with a public rollout expected next month.