Disrupting a Criminal Scam Operation
OpenAI disrupted a Cambodia-based criminal operation using ChatGPT for investment, romance, gambling, and impersonation scams.
OpenAI News·4 Aug
Search signals, briefings, benchmarks and glossary terms.
Search signals, briefings, benchmarks and glossary terms.
Use this view to inspect the underlying evidence corpus. For ranked developments, decision posture and interpretation, use Signals.
Raw feed or Signals?
Raw feed is chronological evidence. Signals ranks and interprets material change.
OpenAI disrupted a Cambodia-based criminal operation using ChatGPT for investment, romance, gambling, and impersonation scams.
Chinese AI researchers are increasingly using X to share their work and recruit talent, contrasting with quieter Western counterparts.
AI-generated viral melodramas are proliferating on social media, driven by creators monetizing clickbait content.
OpenAI outlines a 'full-stack approach' to make advanced AI more capable, affordable, and widely useful, focusing on infrastructure and models.
OpenAI details its safety, security, transparency, and provenance practices for responsible AI governance, particularly in the European context as the EU AI Act progresses.
European executives express optimism regarding AI investments and returns, despite challenges from global AI infrastructure buildout and tightening supply chains.
Dutch insurer Univé adopted ChatGPT Enterprise for its workforce, integrating leadership, governance, and employee-led innovation to scale AI use.
The European Commission's AI Office will begin enforcing the AI Act and new transparency rules on August 2, 2026, for specific AI systems.
China's DeepSeek has released a public beta API for its V4 Flash AI model, citing advancements in agentic capabilities.
Kioxia missed earnings forecasts, signaling a potential slowdown in the AI-driven surge in flash memory prices, impacting data center infrastructure costs.
Leopold Aschenbrenner's Situational Awareness hedge fund sold billions in AI-related tech investments at a discount due to bank collateral demands.
Chipmakers report record earnings, yet investor skepticism and high expectations are causing stock declines, indicating a cooling AI investment climate.
Murata Manufacturing, a key AI component supplier, raised its profit outlook but cautioned that the current surge in AI spending will eventually moderate.
Reported GPT 5.6 price cuts, driven by recursive self-optimization, are claimed to have reduced the cost of GPT 5.4 intelligence by 13x in four months.
Researchers introduced Echoverse, a framework for generating deep, evolving synthetic environments to train computer-use agents at scale.
STEREODISCO framework measures and discovers previously unexamined stereotypical associations and biases in LLMs using semantic differential method.
A research survey reviews methods for Uncertainty Quantification (UQ) in deep learning, focusing on ensemble and approximate Bayesian approaches.
Research proposes a hard-routed mixture-of-experts approach for combining independently trained LoRA adapters in LLMs, preserving original update scales.
SNAC-Pack 2.0 introduces Scaled-Out Surrogate Neural Architecture Codesign, optimizing model architecture for hardware cost, especially FPGAs, beyond just accuracy.
Research proposes a method for continual learning in vision-language models to prevent catastrophic forgetting by preserving semantic geometry.
Open-vocabulary BEV segmentation uses vision-language models for autonomous driving perception beyond training sets, improving real-world robustness.
Research investigates deep generative models' ability to reproduce complex, non-stationary spatial and spatio-temporal data distributions, a key challenge for real-world application.
Research proposes Doubly Robust Functional Representation Learning for longitudinal causal inference, handling irregular time-series data in studies.
Research proposes a neurosymbolic approach to imitation learning combining neural networks for high-dimensional data and symbolic methods for generalization.
Research finds Diffusion Language Models (DLMs) are less robust to input noise and adversarial attacks than autoregressive (AR) models like Llama 3.
Research identifies a structural issue in on-policy self-distillation (OPSD) for LLMs, proposing $\beta$-OPSD as a more stable policy optimization approach.
LightRot introduces a lightweight rotation scheme and hardware architecture for energy-efficient, accurate low-bit large language model inference.
Research presents a theoretical error analysis for Engression, a neural-network-based conditional distribution learning method using the energy score.
Research introduces Dynamic Spectral Filtering (DSF), an operator-centric approach for temporal graph learning, allowing propagation mechanisms to evolve.
Research explores adversarial training's generalization in Reproducing Kernel Hilbert Space, deriving error bounds for robustness levels and sample size.
Researchers propose Attention Heads and MLP Pruning (AMP) to significantly reduce LLM computational costs and accelerate inference for resource-limited deployments.
Research introduces a 'three-number reporting' method for evaluating classifiers by matching evaluation conditions to operational class imbalance for Sentinel-1 wave detection.
FORGE is a new LLM training technique that eliminates the need to materialize gradients, reducing memory usage by fusing optimizer steps.
Research explores Graph Neural Networks for multilevel preconditioning to improve iterative solvers for large, sparse linear systems in scientific computing.
Researchers developed MMC+, an enhanced framework for scalable drift monitoring in medical imaging AI, building on the CheXstray framework.
Researchers introduced Dynamically Scaled Activation Steering (DSAS), a method-agnostic framework for guiding generative model behavior more efficiently.
Research introduces procedural fairness, defined as equal voice and representation within a policy, for multi-agent multi-armed bandits.
New research proposes a theoretical framework for variance-aware baselines and adaptive learning rates to improve reinforcement learning with verifiable rewards (RLVR) in LLMs.
New research introduces strategic classification models where agents' responses to classifiers induce genuine improvements, not just cosmetic feature changes.
Research introduces Conformal Cascade, a multi-tier LLM inference method offering distribution-free accuracy guarantees, addressing miscalibrated confidence scores.
Research suggests that training future AI systems on content generated by a diversity of models, rather than a single model, mitigates knowledge collapse.
Research proposes Quadratic Objective Perturbation (QOP) for differentially private empirical risk minimization, addressing limitations of Linear Objective Perturbation (LOP) with unbounded gradients.
New research shows previous claims of large language models learning effectively from 100% noisy data using RLVR are invalid due to data contamination.
Research proposes a neural network method for anomaly detection in cybersecurity, addressing novel attack evasion without prior anomaly data or distributions.
Research explores continuous-time reinforcement learning for optimal switching problems across multiple regimes, using an exploratory formulation.
Researchers introduced Theseus, a training-free method to transport task-specific parameter updates across large language models with different architectures.
Research characterizes the sparsity and geometry of parameter updates in on-policy distillation, a post-training recipe for language and vision-language models.
A research survey on continual learning for vision-language models (VLMs) addresses catastrophic forgetting in adapting to non-stationary data.
OneShot introduces a neural scoring approach for large-scale retrieval systems, improving ranking accuracy while maintaining indexing efficiency.
Research introduces importance-aware aggregation for federated graph learning (FGL) to improve global model performance under domain shifts across clients.
© 2026 OneBench: AI Insights. All rights reserved.
Evidence before opinion