xAI put Grok inside Microsoft Word, SpaceX raced Google to put AI servers in orbit, and deepfake job fraud just got its first state law: all in one day.
Grok is now inside Microsoft Word, and it's free
xAI launched Grok for Microsoft Word, a free add-in for Microsoft 365 that can turn rough notes into polished documents, rewrite text for clarity, pull in live web research, and connect to email, SharePoint, and Google Drive, all without leaving Word. The same update also brought Grok 4.3 to Amazon Web Services, with a one-million-token context window and the lowest hallucination rate of any frontier model currently available.
Why it matters: If you use Microsoft Office at work or school, AI just landed directly in your writing workflow, no new app, no subscription, no learning curve. This is what "AI at work" actually looks like for most people.
SpaceX wants to run AI from orbit. Google is doing the same.
SpaceX unveiled its AI1 satellite: a solar-powered orbital supercomputer with a 70-metre wingspan that runs AI workloads in space and radiates heat directly into the cold vacuum of space rather than using water-cooled ground facilities. Google is developing a competing 81-satellite cluster called Suncatcher. Both companies are betting that as AI demand for power doubles by the end of the decade, space is a cleaner and more scalable place to put the computing.
Why it matters: AI's energy hunger is becoming one of its biggest political and environmental problems on Earth. Space-based computing is a real long-term bet by the largest players, not just a concept.
White House sets the rules: one national AI policy, states step aside
A June executive order from the White House formally establishes a national AI innovation and security framework, tasking US agencies to lead with a "minimally burdensome" regulatory approach. The order preempts state-level AI laws (meaning individual US states can no longer set their own rules), and directs federal funding toward AI vulnerability detection programmes. Within 30 days of the order, agencies must identify grant funds that can be redirected to AI safety research.
Why it matters: The US is choosing a single, federal approach to AI governance rather than a patchwork of 50 state laws. That makes the US rulebook cleaner for companies, and signals that Washington sees AI leadership as a national security matter, not just a tech trend.
Deepfake job candidates are now mainstream. Colorado just drew a line.
AI-powered fake job applicants (using real-time face and voice cloning to pass video interviews) have become a documented crisis for hiring teams globally. Tools that used to require film-studio budgets now work through a web browser. AI job fraud losses have surged from $90M in 2020 to over $500M in 2024. Colorado's new AI governance law, which took effect this month, is the first US state measure specifically addressing this category of AI-enabled fraud and placing disclosure obligations on employers using AI in hiring.
Why it matters: If you are hiring anyone remotely, or applying for remote roles, the game has changed. Verification steps that worked five years ago no longer work. Colorado's law is the first domino, more states will follow.
India opens its global accelerator for AI startups, and adds AI to every classroom
Applications are open for the IndiaAI Startups Global Programme, Cohort 2, a four-month acceleration programme in partnership with Station F in Paris and HEC Paris, designed to take Indian AI startups to international markets. Separately, India is rolling out a mandatory AI curriculum from Grade 3 across all schools starting the 2026–27 academic year, aligned with the National Education Policy. The IndiaAI Mission's AIKosh platform now hosts over 7,500 open datasets and 273 AI models for public use.
Why it matters: India is investing in AI at two ends simultaneously: the next generation of students and the next generation of startups. If you are building in India, the global door is being held open. If you are a student, AI literacy is now officially in the syllabus.
AI video generation just got noticeably better and faster
xAI released Grok Imagine Video 1.5, now generally available via its Imagine API. The update delivers better motion, more realistic physics, and improved audio in generated videos, all at faster speeds than the previous version. The announcement comes as consumer video AI tools from Google's Veo 3 and OpenAI's Sora 2 also compete for the same space, collectively making high-quality AI video accessible to anyone with a browser.
Why it matters: Content creation (ads, explainer videos, social media) is being permanently reshaped. The baseline for what one person can produce alone is rising every month.
The EU AI Act becomes fully enforceable in 6 weeks
August 2, 2026 is when the EU AI Act becomes fully applicable, meaning companies using AI in Europe face real legal obligations around transparency, data governance, and risk classification for the first time. Rules on prohibited AI practices and general-purpose AI models are already live. The companies that treated this deadline as distant now have six weeks to catch up. Investors are already treating regulatory preparedness as a core valuation factor, calling it a "compliance premium."
Qualcomm eyes a $10B chip deal + Meta cuts 8,000 jobs to go all-in on AI
Qualcomm is in early talks to acquire Tenstorrent (the AI chip company run by chip legend Jim Keller) for between $8 and $10 billion, a move that would put Qualcomm directly in the race for AI infrastructure. Meanwhile, Meta confirmed layoffs of approximately 8,000 employees (roughly 10% of its workforce), while simultaneously reassigning 7,000 others to AI-focused teams. Both stories reflect the same underlying shift: large companies are restructuring their human and financial capital around AI at speed.