A developer named Aniruddha Adak, a final-year computer science student, merged 348 pull requests across dozens of repositories. He shared 11 critical bugs he encountered in modern AI and web frameworks, including cross-…
AI search engines like ChatGPT are reshaping digital intermediation by resolving information needs internally, reducing outbound clicks to just 5.2% of sessions compared to Google's higher referral rate. This shift weake…
Researchers developed the Neuro-Symbolic Knowledge- and Gradient-Guided Reinforcement Learning (KGRL) algorithm, which uses domain knowledge stored in a Datalog knowledge base to prune non-applicable actions in Parametri…
Researchers introduced IQA-T1, a new framework for image quality assessment that combines multimodal large language models with explicit perceptual observations, achieving state-of-the-art performance on seven benchmarks…
A study analyzing public sentiment on AI ethics in education, particularly after ChatGPT's release, found that overall sentiment remains positive but is punctuated by concerns over academic integrity and ethical oversigh…
AutoTrace, a new tool for vulnerability trigger localization, uses layer-by-layer exploration of code property graphs with LLM agents to identify specific trigger points in software code. It achieves a VulnHit rate of 75…
Researchers have introduced Causal Rules and Maximally Specific Causal Relationships (MSCRs) to resolve statistical ambiguity in AI, ensuring consistent predictions by refining causal rules with all statistically relevan…
A new study reveals that intermediate representations from models like CLIP and DINOv2, combined with semantically aligned references, significantly improve the accuracy of synthetic image attribution. The research highl…
Researchers enhanced Monte Carlo Tree Search with dynamic resource allocation strategies, improving AI performance in high-uncertainty games like Jaipur, Lost Cities, and Splendor. The Dynamic Number of Determinizations …
A new theoretical framework called Compression Calculus posits that intelligence is fundamentally about compressing information and recombining those compressed units, challenging current AI approaches that rely on token…
A study on subjective expected utility (SEU) sensitivity reveals that while theoretical identifiability of decision-making parameters is possible, practical recovery at realistic sample sizes remains negligible, with con…
Google researchers introduced DiffusionGemma, a 26-billion parameter speech recognition model that uses discrete diffusion to transcribe speech in parallel, achieving a 6.6% word error rate on LibriSpeech test-clean. The…
Chinese hedge funds that profited from AI stock rallies are reducing exposure, citing concerns the surge may be unsustainable. Shanghai Everlead Capital and Hunjin Capital have trimmed positions in AI-related stocks, whi…
Researchers have developed a method to reduce sycophancy in AI language models by using counterfactual report mediators that separate genuine evidence from external pressures. The approach achieved a perfect resist-and-u…
The European AI Act imposes new transparency, fairness, and human oversight requirements on high-risk AI systems, particularly in algorithmic hiring. Without comprehensive standards yet in place, companies face complianc…
Australia's government, led by Prime Minister Anthony Albanese, announced new regulations on data centers to control their location and environmental impact, and strengthened copyright protections for creatives against u…
South Korea's AI semiconductor boom has sparked debate over how to define and share 'excess' profits, with industry officials questioning the lack of clear standards for what constitutes excess gains and which partners c…
UltraWork launches a hosted AI coding environment at $399/month with no per-token metering, offering curated models like Kimi K2.7 Code and prompt templates for developers. The service targets indie hackers and small tea…
A company is considering adding an emotion tracking feature to its AI smartwatches that would monitor parameters like anxiety, depression, and stress, and is seeking user feedback on the idea.
A group of 26 former Meta employees filed a federal lawsuit in California accusing the company of using biased AI software to monitor emails, keystrokes, and web activity, then unfairly selecting workers on sick leave or…