Today's roundup spans the practical and the profound: AI quietly writing most of one company's code, eye-watering sums chasing AI infrastructure, and scientists peering into how our own brains break habits. Here's what's worth knowing on June 8, 2026 — with honest caveats on each.
AI is now writing most of Anthropic's own code
Anthropic — the company behind Claude — reported that more than 80% of the code merged into its production codebase in May 2026 was authored by its own AI (Anthropic), with humans reviewing and directing the work.
Why it matters: it's a concrete, high-stakes example of AI-assisted development at scale inside a company whose entire business is getting AI right. The caveat: "authored by AI" still means humans set direction, review, and own the outcome — and a frontier AI lab is hardly a typical software team. Treat it as a striking data point, not a universal benchmark. If you want your own AI tools to pull more weight, our guide to writing AI prompts that work is the place to start.
The price of AI: Google's $80 billion infrastructure push
Google announced plans to raise around $80 billion to fund its AI infrastructure buildout, alongside new water commitments for its data centers — including a goal to become water-positive by 2030.
Why it matters: the numbers underline that today's AI race is, at bottom, an infrastructure race — capital, chips, power, and water. The sustainability pledges acknowledge a growing concern about AI's physical footprint. The caveat: fundraising plans and 2030 targets are commitments, not results; the real test is delivery.
Zoom brings AI into the meeting itself
Zoom introduced ZoomMate, a tool that sits inside live meetings and connects decisions made in the conversation to work platforms like Salesforce, Jira, ServiceNow, and Slack, reportedly priced around $20 per user per month.
Why it matters: it's part of a clear trend — AI moving from a separate chatbot into the tools where work already happens, turning "what we decided" into action items automatically. The caveat: in-meeting AI raises real questions about recording, privacy, and accuracy; useful, but worth configuring carefully. (Choosing a chat-and-meetings stack? See our Slack vs Microsoft Teams comparison.)
Your brain has a "break the habit" signal
On the science side, researchers report identifying a brain signal that helps us break old habits and adapt when circumstances suddenly change — a glimpse into the neuroscience of flexibility (via ScienceDaily, June 8).
Why it matters: understanding how the brain switches out of autopilot could eventually inform how we treat compulsive behaviors and learn new routines. The caveat: this is early neuroscience research; from a lab finding to a practical application is a long road.
In nutrition, the source may matter more than the amount
A large, long-term study suggests that for dietary nitrate, where it comes from (for example, vegetables versus processed sources) may matter far more than how much you consume (via ScienceDaily, June 8).
Why it matters: it's a useful reminder that "a nutrient" isn't one thing — context and source change the health story. The caveat: observational studies show associations, not proof of cause; this is information, not dietary advice — talk to a professional for that.
The takeaway
The throughline today: AI is getting embedded — into codebases, infrastructure, and meetings — while science keeps reminding us how much we're still learning about ourselves. Most of it is genuinely useful and genuinely early. Stay curious, keep humans in the loop, and verify the bold claims. For a deeper look at the in-house-AI shift behind today's headlines, see why big tech is building its own AI models.
Sources linked or attributed inline. Company claims are attributed, not endorsed; research-stage and observational findings are flagged as such.
