A research team from the University of Glasgow says it may have solved one of the biggest hurdles standing between today’s 5G and the promised ultra‑fast, ultra‑reliable 6G networks. The problem – electromagnetic interference (EMI) – has plagued early 6G experiments because the tricks used to suppress unwanted noise also end up weakening the desired signal. The Glasgow group turned to a cutting‑edge approach called Reconfigurable Intelligent Surfaces (RIS). Think of RIS as a wall covered in thousands of tiny, programmable antenna‑like elements. Each element can be individually tuned to reflect, focus, or redirect electromagnetic waves. By arranging these elements across a surface, the researchers can steer interference away while simultaneously amplifying the intended 6G signal. Early lab tests show the RIS‑based system can cut EMI without the usual loss of signal strength, a feat that could unlock the massive data rates and ultra‑low latency envisioned for 6G‑enabled applications such as holographic video calls, real‑time remote surgery, and massive Internet‑of‑Things ecosystems. The work, conducted in partnership with industry labs like IMEC and backed by telecom giant Ericsson, marks a significant step toward making next‑generation connectivity a reality.
Read more
Over a recent weekend, users of Anthropic’s Claude AI were startled to find that some of their supposedly private conversations were appearing in public web‑search results. The leak was first spotted by a Reddit user and quickly spread, revealing chats that ranged from political party advice and legal ethics questions to explicit role‑play scenarios. Claude lets people create a shareable link—a “snapshot”—of a specific chat thread. While these links are meant to be shared only with chosen recipients, search‑engine crawlers indexed several of them, making the content searchable to anyone. Anthropic tries to block this by adding a robots.txt directive that tells crawlers not to scrape shared‑chat URLs, a practice they’ve followed since at least September 2025. However, the reality is that keeping pages out of search results is more complicated than a simple file can guarantee. The incident highlights a broader challenge: as generative‑AI tools blend with the open web, the line between private and public data can blur. Users should treat shared links as potentially public, and AI providers need stronger safeguards to honor privacy expectations.
Read more
Cascade, a new data‑driven platform for the building industry, just closed a $3.5 million funding round to help contractors locate and win the right projects. Co‑founders Ana Ferreira and Zia Khan say they grew up watching family members struggle to find work despite their expertise – Ferreira’s mother sold construction supplies, while Khan’s father failed to land enough jobs in Pakistan. Their solution is simple: gather every public grant, permit and bidding portal into one searchable dashboard and use predictive analytics to tell firms which developers are most likely to win a new contract. “We connect that data and say, ‘One of these five developers will probably get the grant – start talking to them now,’” Ferreira explains. The startup’s momentum was boosted after a standout demo at Andreessen Horowitz’s Speedrun program, which opened doors to heavyweight clients that have built JFK and LaGuardia airports, Four Seasons hotels and major data centers. The fresh capital will fund a market push, industry events and new engineering hires. In a landscape where builders must log into dozens of state, city and federal sites to chase a single bridge project, Cascade aims to turn that treasure‑hunt into a streamlined, data‑backed process that lets construction firms focus on what they do best – building.
Read more