https://aiexpert.news/en/article/beyond-patch-summary-ai-assisted-code-review-that-catches-implicit-logic-changes ai|expert en 2026-05-26T06:20:15.462Z IBM Framework Classifies Code Changes at 84% Recall https://aiexpert.news/pt/article/beyond-patch-summary-ai-assisted-code-review-that-catches-implicit-logic-changes ai|expert pt-BR 2026-05-26T06:20:15.462Z Framework IBM Classifica Mudanças de Código com 84% de Recall https://aiexpert.news/es/article/beyond-patch-summary-ai-assisted-code-review-that-catches-implicit-logic-changes ai|expert es 2026-05-26T06:20:15.462Z Marco de Trabajo IBM Clasifica Cambios de Código con 84% de Recall https://aiexpert.news/en/article/forgiving-models-robust-teams-self-generated-replay-stops-catastrophic-forgettin ai|expert en 2026-05-26T05:48:15.397Z Self-Generated Replay Cuts Catastrophic Forgetting in Fine-Tuned Models https://aiexpert.news/pt/article/forgiving-models-robust-teams-self-generated-replay-stops-catastrophic-forgettin ai|expert pt-BR 2026-05-26T05:48:15.397Z Reprodução Autogenerada Reduz o Esquecimento Catastrófico em Modelos Ajustados https://aiexpert.news/es/article/forgiving-models-robust-teams-self-generated-replay-stops-catastrophic-forgettin ai|expert es 2026-05-26T05:48:15.397Z La Reproducción Autogenerada Reduce el Olvido Catastrófico en Modelos Ajustados https://aiexpert.news/en/article/automated-benchmark-auditing-catching-hidden-flaws-in-ai-eval-benchmarks-at-scal ai|expert en 2026-05-26T05:16:15.181Z Stanford Framework Reveals Hidden Flaws in AI Benchmarks https://aiexpert.news/pt/article/automated-benchmark-auditing-catching-hidden-flaws-in-ai-eval-benchmarks-at-scal ai|expert pt-BR 2026-05-26T05:16:15.181Z Framework da Stanford Revela Falhas Ocultas em Benchmarks de IA https://aiexpert.news/es/article/automated-benchmark-auditing-catching-hidden-flaws-in-ai-eval-benchmarks-at-scal ai|expert es 2026-05-26T05:16:15.181Z Marco de Stanford Revela Defectos Ocultos en Benchmarks de IA https://aiexpert.news/en/article/mobilegym-open-agentic-environment-for-verifiable-mobile-gui-agents-at-scale ai|expert en 2026-05-26T04:45:55.143Z MobileGym Solves Mobile-Agent Reproducibility at Scale https://aiexpert.news/pt/article/mobilegym-open-agentic-environment-for-verifiable-mobile-gui-agents-at-scale ai|expert pt-BR 2026-05-26T04:45:55.143Z MobileGym Resolve Reprodutibilidade de Agentes Móveis em Escala https://aiexpert.news/es/article/mobilegym-open-agentic-environment-for-verifiable-mobile-gui-agents-at-scale ai|expert es 2026-05-26T04:45:55.143Z MobileGym Resuelve Reproducibilidad de Agentes Móviles a Escala https://aiexpert.news/en/article/smoke-test-1779763521430 ai|expert en 2026-05-26T03:12:14.985Z Claude Opus 4.7 hits 1M context window at unchanged pricing https://aiexpert.news/pt/article/smoke-test-1779763521430 ai|expert pt-BR 2026-05-26T03:12:14.985Z Claude Opus 4.7 atinge janela de contexto de 1M com preços inalterados https://aiexpert.news/es/article/smoke-test-1779763521430 ai|expert es 2026-05-26T03:12:14.985Z Claude Opus 4.7 alcanza ventana de contexto de 1M con precios sin cambios https://aiexpert.news/en/article/microsoft-mdash-scaling-vulnerability-detection-across-ai-workloads ai|expert en 2026-05-25T20:26:15.689Z Microsoft MDASH Finds 16 Windows Bugs with 100 AI Agents https://aiexpert.news/pt/article/microsoft-mdash-scaling-vulnerability-detection-across-ai-workloads ai|expert pt-BR 2026-05-25T20:26:15.689Z Microsoft MDASH Encontra 16 Bugs do Windows com 100 Agentes IA https://aiexpert.news/es/article/microsoft-mdash-scaling-vulnerability-detection-across-ai-workloads ai|expert es 2026-05-25T20:26:15.689Z Microsoft MDASH Encuentra 16 Bugs de Windows con 100 Agentes IA https://aiexpert.news/en/article/measuring-llm-explanation-quality-narrative-vs-persuasive-framing-in-human-judgm ai|expert en 2026-05-25T06:48:15.306Z Study: AI Narrative Explanations Boost User Trust, Not Accuracy https://aiexpert.news/pt/article/measuring-llm-explanation-quality-narrative-vs-persuasive-framing-in-human-judgm ai|expert pt-BR 2026-05-25T06:48:15.306Z Estudo: IA Aumenta Confiança, Não Precisão do Usuário https://aiexpert.news/es/article/measuring-llm-explanation-quality-narrative-vs-persuasive-framing-in-human-judgm ai|expert es 2026-05-25T06:48:15.306Z Estudio: Las Explicaciones Narrativas de IA Aumentan la Confianza del Usuario, No la Precisión https://aiexpert.news/en/article/from-raw-experience-to-skill-consumption-agent-skill-reuse-study ai|expert en 2026-05-25T05:52:15.543Z Model Scale Fails to Predict Extracted Skill Performance https://aiexpert.news/pt/article/from-raw-experience-to-skill-consumption-agent-skill-reuse-study ai|expert pt-BR 2026-05-25T05:52:15.543Z Escala de Modelo Não Prevê Desempenho de Habilidades Extraídas https://aiexpert.news/es/article/from-raw-experience-to-skill-consumption-agent-skill-reuse-study ai|expert es 2026-05-25T05:52:15.543Z La Escala del Modelo No Predice el Desempeño de Habilidades Extraídas https://aiexpert.news/en/article/developer-scraps-buggy-agentmemory-after-week-in-production-ships-ai-memory-buil ai|expert en 2026-05-25T05:20:15.337Z Five Bugs Killed agentmemory in Seven Days https://aiexpert.news/pt/article/developer-scraps-buggy-agentmemory-after-week-in-production-ships-ai-memory-buil ai|expert pt-BR 2026-05-25T05:20:15.337Z Cinco Bugs Mataram agentmemory em Sete Dias https://aiexpert.news/es/article/developer-scraps-buggy-agentmemory-after-week-in-production-ships-ai-memory-buil ai|expert es 2026-05-25T05:20:15.337Z Cinco Bugs Mataron agentmemory en Siete Días https://aiexpert.news/en/article/llms-as-noisy-channels-shannon-law-explains-scaling-non-monotonicity ai|expert en 2026-05-25T04:48:15.259Z Shannon-Hartley Theorem Explains LLM Quantization Regressions https://aiexpert.news/pt/article/llms-as-noisy-channels-shannon-law-explains-scaling-non-monotonicity ai|expert pt-BR 2026-05-25T04:48:15.259Z Teorema de Shannon-Hartley Explica Regressões de Quantização em LLMs https://aiexpert.news/es/article/llms-as-noisy-channels-shannon-law-explains-scaling-non-monotonicity ai|expert es 2026-05-25T04:48:15.259Z Teorema de Shannon-Hartley Explica Regresiones de Cuantización en LLMs https://aiexpert.news/en/article/complete-mue-hyperparameter-transfer-for-scaling-moe-models ai|expert en 2026-05-25T04:16:15.142Z Complete-muE Lets Teams Transfer Dense Hyperparameters to MoE https://aiexpert.news/pt/article/complete-mue-hyperparameter-transfer-for-scaling-moe-models ai|expert pt-BR 2026-05-25T04:16:15.142Z Complete-muE Permite que Equipes Transfiram Hiperparâmetros Densos para MoE https://aiexpert.news/es/article/complete-mue-hyperparameter-transfer-for-scaling-moe-models ai|expert es 2026-05-25T04:16:15.142Z Complete-muE Permite que los Equipos Transfieran Hiperparámetros Densos a MoE https://aiexpert.news/en/article/skillopt-systematic-optimization-for-self-evolving-agent-skills ai|expert en 2026-05-25T03:45:55.131Z Microsoft's SkillOpt Lifts Agent Accuracy 24 Points via Automated Skill Refinement https://aiexpert.news/pt/article/skillopt-systematic-optimization-for-self-evolving-agent-skills ai|expert pt-BR 2026-05-25T03:45:55.131Z SkillOpt da Microsoft Eleva Precisão de Agentes em 24 Pontos via Refinamento Automático de Skills https://aiexpert.news/es/article/skillopt-systematic-optimization-for-self-evolving-agent-skills ai|expert es 2026-05-25T03:45:55.131Z SkillOpt de Microsoft Eleva la Precisión de Agentes 24 Puntos mediante Refinamiento Automático de Skills https://aiexpert.news/en/article/memaudit-detecting-poisoned-agent-memory-via-causal-attribution ai|expert en 2026-05-25T02:26:15.299Z MemAudit Cuts Memory-Poisoning Attacks to 0% https://aiexpert.news/pt/article/memaudit-detecting-poisoned-agent-memory-via-causal-attribution ai|expert pt-BR 2026-05-25T02:26:15.299Z MemAudit Reduz Ataques de Envenenamento de Memória a 0% https://aiexpert.news/es/article/memaudit-detecting-poisoned-agent-memory-via-causal-attribution ai|expert es 2026-05-25T02:26:15.299Z MemAudit Reduce Ataques de Envenenamiento de Memoria a 0% https://aiexpert.news/en/article/fraud-layered-spvs-and-1-trillion-in-demand-inside-anthropics-pre-ipo-secondary- ai|expert en 2026-05-24T23:12:15.321Z $1 Trillion Chases Anthropic Shares in Fraud-Prone Market https://aiexpert.news/pt/article/fraud-layered-spvs-and-1-trillion-in-demand-inside-anthropics-pre-ipo-secondary- ai|expert pt-BR 2026-05-24T23:12:15.321Z $1 Trilhão Corre Atrás de Ações da Anthropic em Mercado Propenso a Fraudes https://aiexpert.news/es/article/fraud-layered-spvs-and-1-trillion-in-demand-inside-anthropics-pre-ipo-secondary- ai|expert es 2026-05-24T23:12:15.321Z $1 Billón Persigue Acciones de Anthropic en Mercado Propenso a Fraude