Latest AI (Artificial Intelligence) News
Block Lays Off 40% of Staff, Citing AI Efficiency Gains
Financial tech company Block, led by Jack Dorsey, announced layoffs of over 4,000 employees, stating AI tools enable a smaller team to achieve more. Dorsey noted in a shareholder letter that intelligence tools have transformed company operations, with shares soaring over 20% post-announcement.
Experts view this as a potential tipping point for white-collar job displacement.
US Tech Firm's AI Layoffs Spark Concerns for Canadian Workforce
A major US tech firm announced thousands of layoffs to replace workers with AI, amid Canada's slowest economic growth in a decade. Global News reports this as a possible wake-up call, with Canadian industries showing resilience compared to US sectors like finance adopting AI aggressively.
TD Economics notes minimal job displacement in Canada so far.
Trump Directs Federal Agencies to Halt Anthropic AI Use
President Trump ordered all federal agencies to stop using Anthropic's AI after the firm denied unrestricted military access. Anthropic, the only AI on classified military systems, refused terms to prevent mass surveillance or autonomous weapons.
The Pentagon threatened to end the contract by a 5:01 p.m. ET deadline.
University of Texas Wins Gordon Bell Prize for AI Digital Twins in Tsunami Forecasting
A UT Austin-led team earned the 2025 ACM Gordon Bell Prize for AI-powered digital twins forecasting tsunamis on the Cascadia Subduction Zone. Their algorithms achieved a 10-billion-fold speedup, reducing 50 years of supercomputing to seconds.
This advances AI for science in safety, energy, and medicine.
DOE's Genesis Mission to Accelerate Science with Agentic AI
The Department of Energy's Genesis Mission integrates agentic AI into national labs for rapid scientific discovery, handling the full scientific method. Partnerships with Microsoft, Google, Anthropic, OpenAI, and Nvidia provide supercomputing power.
Official Gil envisions an 'internet of science' for fusion and quantum advances.
Google's AI Solves Decade-Old Scientific Problem in 48 Hours
Google's advanced AI system cracked a problem that stumped scientists for over 10 years, completing it in just 48 hours. This breakthrough highlights AI's potential to revolutionize scientific research timelines.
Details on the specific problem remain under wraps in reports.
CHEO's ThinkRare AI Expands Nationally for Rare Disease Detection
The CHEO Research Institute's ThinkRare algorithm, scanning EMRs for undiagnosed rare genetic conditions in children, launches national expansion. It has flagged 41 cases with a 70% diagnosis rate, shortening diagnostic odysseys.
Developed collaboratively, it supports clinicians and reduces repeat tests.
New Benchmarks Test AI's Scientific Reasoning and Discovery Capabilities
OpenAI's FrontierScience and Scientific Discovery Evaluation (SDE) benchmarks assess AI on expert-level science tasks and real-world research scenarios. FutureHouse's LABBench2 evaluates agentic AI from idea to paper, revealing gaps in data handling.
Researchers emphasize benchmarks driving AI progress like ImageNet did.
Canadian AI Minister Seeks OpenAI Clarity After Tumbler Ridge Incident
Canada's AI Minister demands answers from OpenAI CEO on recent changes following the Tumbler Ridge event. Ottawa and BC Premier plan meetings for national AI safety standards.
Concerns balance innovation with public safety.
Debate Emerges: Is Artificial General Intelligence Already Here?
Experts question if AGI exists unrecognized, arguing expectations for groundbreaking discoveries set unrealistically high bars. Current AI excels in patterns but integrates with human efforts in science.
This challenges traditional AGI benchmarks.
2026 IAI Spotlight Highlights AI Innovation and Collaboration
The 2026 IAI event showcased AI's intersection with human well-being and scientific discovery through diverse topics. Breakout sessions emphasized collaborative research advances.
It positions AI as central to future innovations.
AI for Science Benchmarks Push Toward Autonomous Researchers
New tests like LABBench2 gauge if LLMs can complete full scientific workflows, from literature search to experiments. Leading models succeed in searches but falter in complex data tasks.
Progress hinges on improved information retrieval.