Latest AI (Artificial Intelligence) News

đź“…February 19, 2026 at 1:00 AM
AI advances in drug discovery and scientific research continue accelerating, while artists face significant income losses from generative AI and safety concerns grow about uncontrolled AI development.
1

Helen Toner Warns of 'Hindenburg-Style' AI Risks as Global Race Accelerates

CSET expert Helen Toner discussed on CNN the accelerating global AI race and growing challenges posed by China's rapid AI advancement and the shift toward autonomous AI agentsSource 1. Toner emphasized that these powerful systems could fail unpredictably or have unintended consequences, with engineers themselves raising alarms about potential serious harmSource 1.

2

First AI-Discovered Drug Advances Through Human Clinical Trials

Researchers at Isomorphic Labs, a Google DeepMind spinoff, are using new versions of AlphaFold to study previously undruggable proteinsSource 2. The company has developed rentosertib, a drug discovered entirely by AI systems, which appears safe and effective in small human trials and could become the first AI-discovered drug on the market if it passes larger clinical trialsSource 2.

3

Artists Face Up to 24% Income Decline Due to AI, UNESCO Reports

A new UNESCO report warns that music creators could see revenues fall by 24 percent while audiovisual workers may lose 21 percent of income by 2028 due to expanding AI-generated contentSource 3. The report notes that these disruptions are occurring faster than current policy responses can address, exacerbating inequalities among millions of cultural workersSource 3.

4

€60 Million LIGAND-AI Project Launches to Revolutionize Drug Discovery

An international consortium of 18 partners across nine countries launched a five-year project to train AI models for predicting drug-like molecules targeting thousands of human proteinsSource 4. The project will test up to 0.5 million compounds across approximately 2,000 human proteins, generating high-quality datasets that could expand drug discovery beyond the current 700 proteins typically usedSource 4.

5

EDEN AI System Uses Evolutionary Models to Design New Gene Therapies

A collaboration between Nvidia, Microsoft, Basecamp Research, and University of Pennsylvania researchers developed EDEN, an AI system trained on genetic data from a million species to design programmable therapiesSource 5. The system demonstrated the ability to design peptides with 97% efficacy against antibiotic-resistant infections and CAR-T lymphocytes with 90% efficacy against tumor cells in laboratory trialsSource 5.

6

Google DeepMind Partners with India on Frontier AI for Science and Education

Google DeepMind established a new partnership with Indian government bodies and local institutions to provide access to frontier AI models for science and educationSource 8. The initiative includes powering innovation hubs with generative AI assistants and collaborating on AI-powered climate solutionsSource 8.

7

Stanford AI Predicts Disease Risk Using Single Night of Sleep Data

Stanford researchers developed an AI system that analyzes physiological signals from one night of sleep to predict future risks of conditions including cancer, dementia, and heart diseaseSource 6. The system successfully identifies hidden patterns across brain, heart, and breathing signals that doctors have largely overlookedSource 6.

8

Breakthrough: Quantum Computers Could Scale to Millions of Qubits

Stanford researchers created miniature optical cavities that efficiently collect light from individual atoms, allowing multiple quantum bits to be read simultaneouslySource 6. Demonstrated working arrays with dozens and even hundreds of cavities could eventually support massive quantum networks with millions of qubitsSource 6.

9

Study Questions Whether AI Advances Benefit Scientific Fields as a Whole

Recent research suggests AI tools help individual scientists make progress but may not be advancing entire scientific fields proportionallySource 7. Researcher Steven Salzberg noted that while AlphaFold 2 was a genuine breakthrough, some newer AI tools for predicting genetic impacts create more work than value for scientistsSource 7.

10

Harvard Researchers Create Ultra-Thin Quantum Metasurface for Scalable Networks

Harvard scientists developed a groundbreaking metasurface thinner than a human hair that can replace bulky optical components in quantum computingSource 6. Using graph theory to simplify design, the metasurface can generate entangled photons and perform quantum operations on a single chip, advancing room-temperature quantum technologySource 6.