Artificial Intelligence

Artificial Intelligence•Jul 22, 2026

This Open-Source AI Tries to Reject Bad Drug Ideas Before the Lab Does

PeptiVerse predicts several make-or-break properties of peptide drug candidates before researchers synthesize them, potentially saving time and money. The open-source system performed strongly on curated benchmarks, but its predictions are filters, not experimental proof that a molecule is safe or effective.

Artificial Intelligence•Jul 21, 2026

The Web Was Built to Keep Websites Apart. AI Agents Are Breaking That Wall

AI browser agents can read and act across websites where users are already logged in. Researchers found that this convenience may weaken a core browser-security boundary, although the demonstrated attack required specific conditions and does not prove every agentic browser is unsafe.

Artificial Intelligence•Jul 20, 2026

AI Is Not Just Writing Code. It Is Rebuilding the Computer Around Us

Coding agents now fix issues and open pull requests, browsers act across websites, and operating systems are becoming platforms for local models. The real opportunity is not replacing engineers, but letting disciplined teams test more ideas and build systems that were previously too expensive to attempt.

Artificial Intelligence•Jul 20, 2026

TSMC Is Spending $265 Billion on US AI Chips. Its Hardest Problem May Be Finding Enough Builders

TSMC says AI-chip demand remains strong enough to justify a record expansion in Arizona. Money and customers are available, but construction labour, specialist skills and the complexity of duplicating Taiwan’s chip ecosystem may determine how quickly the factories actually arrive.

Artificial Intelligence•Jul 19, 2026

Apple Just Beat Nvidia Without Winning the AI Arms Race

Apple briefly reclaimed the title of the world’s most valuable company from Nvidia. The reversal suggests investors may be looking beyond the biggest chipmaker, but it does not mean Apple has solved its AI challenges.

Artificial Intelligence•Jul 19, 2026

AI Helped Make a T. Rex Leather Handbag. The Dinosaur Part Is More Complicated

Scientists used AI-assisted protein reconstruction and engineered cells to grow a leather-like material inspired by fragmented T. rex collagen. The handbag is real, but calling it resurrected dinosaur skin stretches the science much further than the evidence allows.

Artificial Intelligence•Jul 18, 2026

China’s Kimi K3 Just Shook the AI Race, but Its 2.8 Trillion Parameters Come With a Catch

Moonshot AI’s enormous open-weight model is winning praise for coding, vision and a million-token context window. Yet its size, hardware demands and company-run benchmarks show why the headline number is only part of the story.

Artificial Intelligence•Jul 18, 2026

AI Found a Display Recipe Humans Missed, Making Quantum-Dot LEDs Last More Than 40 Times Longer

Researchers used machine learning to predict the solvent properties needed to pack quantum dots into smoother, denser films. A mixed-solvent recipe derived from the model roughly doubled QLED efficiency and extended operational lifetime more than fortyfold, although the result remains a laboratory process rather than a commercial display.

Artificial Intelligence•Jul 16, 2026

xAI Is Suing a Grok User Over Sexual Deepfakes, Raising a New Question: Who Is Responsible?

xAI has taken the unusual step of suing a user it accuses of bypassing Grok’s safeguards to generate non-consensual sexual deepfakes involving adults and minors. The case may establish a new enforcement path for AI companies, while intensifying scrutiny of why the system could allegedly be misused at scale.

Artificial Intelligence•Jul 15, 2026

MIT Students Built a Working Jet Engine With AI, Then Found Where AI Breaks

MIT’s JARVIS Challenge asked seven student teams to design and hot-fire small gas-turbine engines with frontier AI copilots. A winning engine generated net thrust, but hallucinations, weak physical understanding and manufacturing delays showed that expert judgment—not the model—remained decisive.

Artificial Intelligence•Jul 15, 2026

KAIST’s Temporary ‘Jailbreak’ Makes Personalized AI Fine-Tuning Safer

KAIST researchers introduced Buffer-and-Reinforce, a two-stage fine-tuning method that temporarily isolates a language model while it learns private or company data, then strengthens its safeguards. Tests reduced harmful-response rates without materially sacrificing customization, but the evidence remains limited to controlled model evaluations.

Artificial Intelligence•Jul 14, 2026

AI Helps Break an 80-Year-Old Erdős Mathematics Barrier

A general-purpose reasoning model produced a new construction for Paul Erdős’s planar unit-distance problem. It does not solve the entire question, but overturns a long-standing assumption and demonstrates a new form of human–AI research collaboration.