Google's personal AI agent rolls out, Microsoft enters the model race, and AI-generated fake job applicants are now a quantified hiring crisis.
Google's always-on AI agent Gemini Spark begins rolling out
Gemini Spark (Google's 24/7 personal AI agent) started reaching Google AI Ultra subscribers in the US this week. Unlike a chatbot you open and close, Spark runs continuously in the background on your phone or laptop, taking actions on your behalf even when your screen is off. It can book appointments, manage files, draft responses, and coordinate other AI tools working in parallel.
Why it matters: This is the shift from AI as a tool you consult to AI as a partner that acts without being asked each time. The way most people interact with their devices is about to feel very different.
Microsoft enters the AI model race with MAI-Code and MAI-Thinking
At its Build developer conference, Microsoft unveiled its first in-house AI models: MAI-Code-1-Flash, built for coding tasks, and MAI-Thinking-1, a reasoning model for complex problem-solving. Both are designed to reduce Microsoft's dependence on OpenAI and offer developers cheaper alternatives when building AI-powered applications.
Why it matters: When a company that co-funded OpenAI starts building its own competing models, it signals that the AI model market is maturing fast. More competition almost always means lower costs and more choice for anyone building with these tools.
OpenAI adds an Ads Manager to ChatGPT, and real-time audio models
OpenAI launched a self-serve Ads Manager inside ChatGPT, allowing businesses to target users with advertising and measurement controls for the first time. In a separate move, three new real-time audio models went live for conversational agents, live translation, and transcription, significantly improving what developers can build with voice-based AI.
Why it matters: ChatGPT adding ads changes the economics of who pays for what. The free tier could become ad-supported, broadening access, but it also raises questions about whether ad revenue influences what ChatGPT recommends.
One in three job applicants may now be AI-generated deepfakes
Security firm Pindrop analysed 300 job applicant profiles and found over a third were entirely fabricated: AI-generated resumes, fake work histories, and deepfake video interviews capable of passing initial HR screening in real time. Gartner projects that 1 in 4 candidate profiles globally will be fake by 2028. The US DEFIANCE Act, which allows deepfake victims to sue creators and platforms, has now passed into law.
Why it matters: Hiring processes built for human applicants are being systematically exploited. If you hire people, your screening likely needs updating. If you're job-hunting, being verifiably human is becoming a genuine competitive edge.
India's top IT firms hired 7,000 fewer people, and new AI content rules are coming
Net hiring at India's top five IT companies dropped by around 7,000 in the financial year ended March 2026, compared to a five-year average of 230,000 gross hires annually. At the same time, India's 2026 IT Rules amendment introduced a new legal category called Synthetically Generated Information (SGI), requiring platforms to clearly disclose whenever audio, video, or images are AI-generated, a move echoing the EU's own deepfake labeling push.
Why it matters: India is building one of the world's largest AI infrastructure programmes while its biggest employment sector slows hiring. The SGI rules are a practical step forward on safety, but enforcement and public awareness will take time to catch up.
EU deepfake labeling enforcement arrives in 56 days
The EU AI Act's Article 50 (which requires all AI-generated audio, video, and images to carry machine-readable labels) takes effect on August 2, 2026. The EU's Code of Practice on AI-Generated Content is being finalised right now. This is the first major AI enforcement deadline with real penalties attached. Any platform operating in the EU must comply or face fines. India's own SGI framework is watching closely to decide whether to follow the same model.
AI literacy gap grows + Anthropic closes a $65B funding round
84% of US students use AI tools regularly, but only 16% of high school leaders say all their students get formal AI literacy education, and 61% of elementary teachers say their students cannot tell AI-generated content from human-made content. Separately, Anthropic closed a $65B Series H at a near-$965B valuation, crossing $47B in annual run-rate revenue and briefly overtaking OpenAI as the most valuable private AI company.