Switzerland's new open-source AI model Apertus tackles what kills 95% of AI projects: lack of transparency. Full visibility into code and training data promises to solve bias, compliance headaches, and corporate trust issues.
Warp challenges "hands-off" agent coding with human-in-the-loop approach, charging $200/month for enterprise tier. While competitors push automation, Warp bets on oversight—real-time code review and mid-flight steering as agents work.
Anthropic's $1.5B settlement with authors establishes first pricing benchmark for AI training data—$3,000 per pirated book. Court ruling split training rights from acquisition methods, reshaping how tech giants source content.
A new artificial intelligence system from China's Shandong First Medical University helps scientists understand how genes turn on and off. Called TRAPT, it maps gene control with record-breaking accuracy.
Led by Professor Li Chunquan, the research team built TRAPT to track proteins that regulate genes - a process central to health and disease. The system combines massive datasets with AI to spot which proteins control specific genes.
Think of your genes as light switches. Different cells need different genes "switched on" at different times. A heart cell needs different genes active than a brain cell does. Proteins called transcriptional regulators (TRs) flip these genetic switches.
Finding which TRs control which genes has stumped scientists for years. Earlier tools gave rough guesses - like using a paper map instead of GPS. TRAPT changes this by analyzing over 20,000 datasets showing where TRs bind to DNA.
The system works in two steps. First, it learns where TRs typically bind across all DNA. Then it sees how these patterns change in specific cells or conditions. This two-step approach helps TRAPT make better predictions about gene control.
Tests show TRAPT beats existing tools handily. It spots the right regulatory proteins 13% more often than the next best method. For some types of regulators, it works up to 200% better than current tools.
Credit: Shandong First Medical University
The team proved TRAPT's real-world value in breast cancer cells. The system found key proteins that drive breast cancer growth, including ESR1 and FOXA1. It also revealed new connections that could help scientists understand the disease better.
TRAPT could help researchers study many diseases beyond cancer. It acts like a universal translator for gene regulation, helping scientists read the complex code cells use to control their genes.
The Shandong team made TRAPT free for scientists worldwide. Researchers can use it through a website or download it to their computers.
"Traditional methods miss a lot of important information about gene control," explains Professor Li. "TRAPT combines AI with comprehensive data to give us a much clearer picture. It's like switching from a paper map to a GPS system for finding gene regulators."
The system especially helps track three types of regulatory proteins:
Transcription factors that bind directly to DNA
Transcription co-factors that help other proteins bind
Chromatin regulators that control DNA packaging
This matters because problems with gene regulation appear in many diseases. Cancer often involves genes turned on when they should be off, or vice versa. Better understanding of gene control could lead to more targeted treatments.
The researchers tested TRAPT on 570 different datasets where scientists had turned specific genes on or off. The system consistently identified the right regulatory proteins, beating other tools across multiple measures.
The team also used TRAPT to study breast cancer, where it correctly spotted proteins known to drive tumor growth. This shows the system can help cancer researchers find important gene regulators.
"We designed TRAPT to handle the complexity of gene regulation," says Professor Li. "It processes huge amounts of data to find patterns that other tools miss."
The system runs on powerful AI techniques including:
Graph neural networks that map connections between proteins
Knowledge distillation that helps the system learn efficiently
Deep learning that spots complex patterns in data
Beyond the tool itself, the team's work provides a valuable resource for other scientists. They collected and organized thousands of datasets about gene regulation, creating one of the field's largest databases.
Why this matters:
TRAPT gives scientists an unprecedented view of gene control, which could speed up research on cancer, heart disease, and other conditions
By making both the tool and data freely available, the Shandong team helps researchers worldwide study gene regulation more effectively
Bilingual tech journalist slicing through AI noise at implicator.ai. Decodes digital culture with a ruthless Gen Z lens—fast, sharp, relentlessly curious. Bridges Silicon Valley's marble boardrooms, hunting who tech really serves.
Students embrace AI faster than schools can write rules. While 85% use AI for coursework, institutions stall on policy—and tech giants step in with billions in training programs to fill the vacuum. The question: who gets to define learning standards?
First survey of 283 AI benchmarks exposes systematic flaws undermining evaluation: data contamination inflating scores, cultural biases creating unfair assessments, missing process evaluation. The measurement crisis threatens deployment decisions.
Tech giants spent billions upgrading Siri, Alexa, and Google Assistant with AI. Americans still use them for weather checks and timers—exactly like 2018. Fresh YouGov data reveals why the utility gap persists.
A new benchmark testing whether AI models will sacrifice themselves for human safety reveals a troubling pattern: the most advanced systems show the weakest alignment. GPT-5 ranks last while Gemini leads in life-or-death scenarios.