News archive.
Earlier AI news and analysis, kept for reference. Current coverage lives in News & Analysis.
Soccer coaches now use AI to break tactical rules that work
Computer scientists at KU Leuven have uncovered hidden patterns in soccer using AI and data analytics, challenging long-held assumptions about how the game should be played. Many insights hitting pitches today trace back to their work.
McDonald's AI drive-thru hits 90% accuracy in 5-location test
McDonald's ArchIQ system processed over 1 million drive-thru orders with 90% completion without staff intervention. The Google-backed tool also monitors restaurant operations—but the company's previous IBM pilot failed after customer complaints.
Apple's Siri now runs on Google's Gemini, locked out of China and EU
Apple rebuilt Siri to hold real conversations and tap your photos, mail and the web—but it relies on Google's models and skips most of the world. Here's who gets it and when.
Your AI agents fail without clean data, Xebia CTO warns
Niels Zeilemaker, global CTO at Xebia, explains why agents misinterpret data and join fields that should stay separate. The fix starts with data catalogues designed for machines, not just humans.
Visa lets ChatGPT buy things without you clicking once
Visa connected its payment network to ChatGPT, letting AI agents pick products and pay across any retailer. Retail operations now need to prepare for transactions that skip websites entirely.
One typo in Linux kernel allows unprivileged users to gain root access
CVE-2026-53111, caused by a single missing character, lets attackers bypass sandbox defenses and escalate privileges. The bug was patched in February; here's what you need to know.
Microsoft patches 0-day after researcher dispute escalates
Microsoft fixed a Windows vulnerability disclosed by security researcher Nightmare Eclipse, but left others unpatched. The dispute raises questions about disclosure timelines and vendor accountability.
Deezer scans your Spotify playlists for AI music—for free
Deezer's AI detection tool now works across 20 streaming services. Upload your playlists to see which tracks are synthetic, no subscription required.
Anthropic Admits Claude Fable's Hidden Guardrails, Commits to Transparency
Anthropic apologized for silently throttling Claude Fable 5 with invisible safety restrictions on distillation queries. The company will now openly redirect restricted requests to an older model instead.
AI memory tools make models ignore facts to please users
Writer researchers found that memory systems designed to personalize AI assistants can backfire, making models sycophantic and less accurate. Here's what the tests showed.
Top 1% of firms spend $7,500 per employee monthly on AI
The most AI-obsessed companies are burning through $7,500 monthly per employee on compute and tokens, per Ramp AI Index. That's still below engineer salaries, but spending is accelerating fast.
xAI Engineer Sues Over Firing After Grok Safety Warnings
Devin Kim claims he was terminated for raising AI safety concerns about Grok ahead of SpaceX's IPO. The lawsuit details the chatbot's documented failures, from Hitler comparisons to nonconsensual imagery.