Full-text search across 12774 articles. Combine with topic and date filters; results sorted by relevance.
Hackers are exploiting lesser-known languages to bypass AI safety mechanisms, challenging tech makers to adapt. The technique targets gaps in AI training datasets, which are typically centered around widely spoken langua…
OpenAI is merging its research and safety departments under a single executive, a restructuring that raises concerns about whether AI safety will be prioritized adequately. The move follows an industry trend toward integ…
Constantin Koehncke, CEO of Image Line, is integrating AI features like stem separation and a chatbot named Gopher into FL Studio, democratizing music production by lowering technical barriers. The company's AI-driven ap…
Finance teams are using ChatGPT Work to generate Monthly Business Reviews and reporting packs in minutes instead of hours, freeing professionals for strategic tasks. The AI tool processes real work inputs to produce comp…
OpenAI is in early discussions to give the US government a 5% stake, CEO Sam Altman said, aiming to align AI benefits with national interest and improve relations with the Trump administration.
Researchers developed a semantic-aligned brain network framework that integrates semantics from large language models into brain disease diagnosis, improving stability and interpretability. The method uses multi-scale hy…
SOLiD, a framework for detecting deceptive AI responses, reduces undetected deception to 14% in models with 405 billion parameters, down from 34% in 1-billion-parameter models, while maintaining a 99% true positive rate.…
Researchers have developed a novel AI method that uses safe reference policies as probabilistic regulators to balance exploration and safety, providing finite-sample guarantees and handling non-monotonic bounded loss fun…
AI enthusiasm has driven stock market gains despite a sluggish U.S. economy, creating a divergence between tech stock performance and broader economic indicators. Investors are betting on AI's transformative potential ac…
A veteran investor has amassed dominant control over the AI sector through massive investments in startups and projects, raising concerns among industry insiders about stifled competition and homogenized innovation. The …
Researchers have identified three core principles—divisibility, transferability, and stability—that govern compositional generalization in AI models, revealing that linear structures in neural representations are essenti…
Researchers introduced PRECEDE, a precedent-guided AI approach for drug redesign that mitigates side effects while preserving therapeutic benefits, using large language models coordinated with human oversight to ensure t…
AI agents are being evaluated for personal negotiations, but a new benchmark called SovereignNegotiation-Bench reveals that high agreement rates often come at the cost of user privacy and utility. The benchmark emphasize…
OpenClaw's GLM-5 inference optimization study found that adjusting parameters like chunked prefill size and request concurrency improved throughput and reduced latency, cutting serving costs by 10.4% per request and 9.6%…
Researchers introduced TREK (Teacher-Routed Exploration via Forward KL), a new AI training method that enhances learning through unconventional exploration strategies. TREK significantly improved performance on mathemati…
Researchers have developed STREAM, a new AI model that uses a diffusion transformer architecture to generate dance motions from both text and music cues, overcoming the modality collapse problem where musical rhythms ove…
Compass, a new AI methodology for prostate cancer detection, analyzes 2D micro-ultrasound studies using a multi-view approach that mimics clinical practice by integrating rotational video sweeps, outperforming traditiona…
The AI market is projected to grow from $189 billion to $4.8 trillion by 2033, but the future of AI lies in small, energy-efficient models rather than power-hungry large language models like GPT-4, which consume 50-60 GW…
A new AI model called continuous-query limited memory language model (CO-LMLM) combines minimal cost with high factual precision by storing knowledge in an external knowledge base and fetching it as needed. At 360 millio…
Researchers introduced MIRA-Math, a benchmark that tests AI models by withholding a single atomic fact in mathematical problems, requiring models to request the missing information in natural language under a strict budg…