https://ainews.lumi-systems.io/en/2026/06/15/harnessx-automated-optimization-of-agent-runtime-environments/ 2026-07-04T10:07:29+00:00 https://ainews.lumi-systems.io/en/2026/06/15/security-mechanisms-of-ai-agents-exploitable-as-dos-weapons/ 2026-07-04T10:07:28+00:00 https://ainews.lumi-systems.io/en/2026/06/15/ai-security-systems-as-dos-targets-poisoned-documents-cripple-guardrails/ 2026-07-04T10:07:26+00:00 https://ainews.lumi-systems.io/en/2026/06/15/litellm-chained-security-flaws-enable-system-takeover/ 2026-07-04T10:07:25+00:00 https://ainews.lumi-systems.io/en/2026/06/16/fastcontext-specialized-agents-for-efficient-code-repository-exploration/ 2026-07-04T10:07:24+00:00 https://ainews.lumi-systems.io/en/2026/06/16/tangram-static-kv-cache-compression-for-faster-multi-turn-llm-serving/ 2026-07-04T10:07:23+00:00 https://ainews.lumi-systems.io/en/2026/06/16/ai-in-product-development-from-isolated-tool-to-integrated-lifecycle-system/ 2026-07-04T10:07:22+00:00 https://ainews.lumi-systems.io/en/2026/06/16/post-training-recipes-at-the-frontier-from-instructgpt-to-mopd-specialist-systems/ 2026-07-04T10:07:20+00:00 https://ainews.lumi-systems.io/en/2026/06/16/p-eagle-parallel-speculation-for-faster-llm-inference-on-aws-sagemaker/ 2026-07-04T10:07:19+00:00 https://ainews.lumi-systems.io/en/2026/06/17/amazon-bedrock-invokeguardrailchecks-api-for-agent-based-applications/ 2026-07-04T10:07:18+00:00 https://ainews.lumi-systems.io/en/2026/06/17/loopcoder-v2-two-loops-as-the-optimum-for-efficient-model-computation-in-programming/ 2026-07-04T10:07:17+00:00 https://ainews.lumi-systems.io/en/2026/06/17/procua-sft-automatically-generated-training-data-for-desktop-agents/ 2026-07-04T10:07:16+00:00 https://ainews.lumi-systems.io/en/2026/06/17/ai-integration-fails-against-outdated-it-architectures/ 2026-07-04T10:07:14+00:00 https://ainews.lumi-systems.io/en/2026/06/17/144-mastra-npm-packages-compromised-via-hijacked-contributor-account/ 2026-07-04T10:07:13+00:00 https://ainews.lumi-systems.io/en/2026/06/17/bucket-squatting-gap-in-vertex-ai-sdk-enabled-model-hijacking/ 2026-07-04T10:07:12+00:00 https://ainews.lumi-systems.io/en/2026/06/17/estonia-plans-digital-identities-for-ai-agents-with-defined-rights/ 2026-07-04T10:07:11+00:00 https://ainews.lumi-systems.io/en/2026/06/17/free-code-production-overturns-the-economics-of-software-development/ 2026-07-04T10:07:10+00:00 https://ainews.lumi-systems.io/en/2026/06/18/glm-5-2-chinese-open-weights-model-with-753-billion-parameters/ 2026-07-04T10:07:09+00:00 https://ainews.lumi-systems.io/en/2026/06/19/efficientrollout-self-speculative-decoding-for-faster-rl-rollouts/ 2026-07-04T10:07:08+00:00 https://ainews.lumi-systems.io/en/2026/06/19/reves-iterative-training-for-more-efficient-test-time-scaling-in-llms/ 2026-07-04T10:07:07+00:00 https://ainews.lumi-systems.io/en/2026/06/19/vertex-ai-sdk-vulnerability-enabled-model-hijacking-on-google-cloud/ 2026-07-04T10:07:06+00:00 https://ainews.lumi-systems.io/en/2026/06/19/microsoft-warns-of-remote-code-execution-through-web-enabled-ai-agents/ 2026-07-04T10:07:04+00:00 https://ainews.lumi-systems.io/en/2026/06/19/uniform-fp4-new-4-bit-training-method-for-llms-reduces-systematic-errors/ 2026-07-04T10:07:03+00:00 https://ainews.lumi-systems.io/en/2026/06/19/ai-agents-as-new-identities-governance-gaps-in-the-enterprise/ 2026-07-04T10:07:02+00:00 https://ainews.lumi-systems.io/en/2026/06/19/autojack-ai-browsing-agents-exploited-for-code-execution-on-host-systems/ 2026-07-04T10:07:01+00:00 https://ainews.lumi-systems.io/en/2026/06/21/auggie-cli-ai-coding-agent-in-terminal-workflow/ 2026-07-04T10:07:00+00:00 https://ainews.lumi-systems.io/en/2026/06/22/harness-engineering-control-mechanisms-for-reliable-ai-agents/ 2026-07-04T10:06:59+00:00 https://ainews.lumi-systems.io/en/2026/06/22/aws-continuum-automates-vulnerability-handling-with-ai-driven-validation/ 2026-07-04T10:06:58+00:00 https://ainews.lumi-systems.io/en/2026/06/22/microsoft-fixes-vulnerability-in-autogen-studio-for-code-execution/ 2026-07-04T10:06:57+00:00 https://ainews.lumi-systems.io/en/2026/06/23/github-actions-security-hardening-against-pull-request-target-abuse-starting-in-version-7/ 2026-07-04T10:06:55+00:00 https://ainews.lumi-systems.io/en/2026/06/23/evoembedding-context-dependent-embeddings-for-long-sequences/ 2026-07-04T10:06:53+00:00 https://ainews.lumi-systems.io/en/2026/06/23/github-restricts-actions-checkout-against-pwn-request-attacks/ 2026-07-04T10:06:52+00:00 https://ainews.lumi-systems.io/en/2026/06/23/parallelkernelbench-frontier-llms-still-struggling-with-fast-multi-gpu-kernels/ 2026-07-04T10:06:51+00:00 https://ainews.lumi-systems.io/en/2026/06/24/naturebench-how-far-coding-agents-really-get-on-scientific-tasks/ 2026-07-04T10:06:50+00:00 https://ainews.lumi-systems.io/en/2026/06/24/edv-framework-reduces-error-accumulation-in-self-learning-llm-agents/ 2026-07-04T10:06:48+00:00 https://ainews.lumi-systems.io/en/2026/06/24/ci-cd-vulnerability-cordyceps-threatens-github-repositories-via-supply-chain-attacks/ 2026-07-04T10:06:47+00:00 https://ainews.lumi-systems.io/en/2026/06/25/github-blocks-automatic-code-checkouts-from-forked-pull-requests/ 2026-07-04T10:06:46+00:00 https://ainews.lumi-systems.io/en/2026/06/25/agentic-overlays-transforming-rest-apis-in-agent-to-agent-communication/ 2026-07-04T10:06:45+00:00 https://ainews.lumi-systems.io/en/2026/06/26/tool-calling-failures-under-schema-constraints-in-open-weight-llms/ 2026-07-04T10:06:44+00:00 https://ainews.lumi-systems.io/en/2026/06/26/language-compression-in-llms-output-optimization-saves-costs-input-reduction-increases-them/ 2026-07-04T10:06:43+00:00 https://ainews.lumi-systems.io/en/2026/06/26/viq-discrete-visual-representations-at-arbitrary-resolution/ 2026-07-04T10:06:41+00:00 https://ainews.lumi-systems.io/en/2026/06/26/openbiorq-benchmark-for-agent-based-biomedical-research-questions/ 2026-07-04T10:06:40+00:00 https://ainews.lumi-systems.io/en/2026/06/26/openbiorq-benchmark-for-agentic-ai-models-in-biomedical-research-questions/ 2026-07-04T10:06:39+00:00 https://ainews.lumi-systems.io/en/2026/06/26/ai-tools-in-software-development-security-gaps-despite-compliance-efforts/ 2026-07-04T10:06:38+00:00 https://ainews.lumi-systems.io/en/2026/06/26/gauntletbench-new-benchmark-reveals-limitations-of-ai-agents/ 2026-07-04T10:06:37+00:00 https://ainews.lumi-systems.io/en/2026/06/26/stripe-scales-compliance-with-ai-agents-26-percent-faster-with-full-control/ 2026-07-04T10:06:35+00:00 https://ainews.lumi-systems.io/en/2026/06/26/infokv-entropy-based-kv-cache-compression-for-long-reasoning-sequences/ 2026-07-04T10:06:34+00:00 https://ainews.lumi-systems.io/en/2026/06/27/github-repository-trick-deceives-ai-agents-into-executing-malware/ 2026-07-04T10:06:33+00:00 https://ainews.lumi-systems.io/en/2026/06/28/one-quarter-of-german-companies-have-no-control-over-ai-development-tools/ 2026-07-04T10:06:32+00:00 https://ainews.lumi-systems.io/en/2026/06/28/integrating-local-language-models-into-production-from-ollama-to-production-ready-code/ 2026-07-04T10:06:31+00:00 https://ainews.lumi-systems.io/en/2026/06/29/ai-generated-code-projects-show-systematic-security-deficiencies/ 2026-07-04T10:06:30+00:00 https://ainews.lumi-systems.io/en/2026/06/29/ornith-1-0-open-source-model-for-agent-driven-software-development/ 2026-07-04T10:06:29+00:00 https://ainews.lumi-systems.io/en/2026/06/29/multi-tenant-llm-analytics-with-row-level-security-par-technologys-architecture-on-aws/ 2026-07-04T10:06:27+00:00 https://ainews.lumi-systems.io/en/2026/06/30/vision-ai-agents-synthetic-data-and-fine-tuning-for-higher-accuracy/ 2026-07-04T10:06:26+00:00 https://ainews.lumi-systems.io/2026/05/10/hugging-face-co-founder-says-qwen-3-6-27b-running-on-airplane-mode-is-close-to-latest-opus-in-claude-code/ 2026-07-04T10:06:25+00:00 https://ainews.lumi-systems.io/2026/05/20/kpmg-integrates-claude-across-its-core-business-and-workforce-of-more-than-276000-in-strategic-alliance/ 2026-07-04T10:06:24+00:00 https://ainews.lumi-systems.io/2026/05/27/chief-of-it-brief-woche-22-2026-ai-omnibus-ghost-cms-cve-nis-2-verantwortung/ 2026-07-04T10:06:23+00:00 https://ainews.lumi-systems.io/2026/05/31/harness-design-for-long-running-application-development-2/ 2026-07-04T10:06:21+00:00 https://ainews.lumi-systems.io/2026/05/31/harness-design-for-long-running-application-development/ 2026-07-04T10:06:20+00:00 https://ainews.lumi-systems.io/2026/05/31/quantifying-infrastructure-noise-in-agentic-coding-evals-2/ 2026-07-04T10:06:19+00:00 https://ainews.lumi-systems.io/2026/05/31/project-glasswing-an-initial-update-2/ 2026-07-04T10:06:18+00:00 https://ainews.lumi-systems.io/2026/05/31/streamline-external-access-to-amazon-sagemaker-mlflow-using-a-rest-api-proxy/ 2026-07-04T10:05:46+00:00 https://ainews.lumi-systems.io/2026/05/31/evaluating-deep-agents-using-langsmith-on-aws/ 2026-07-04T10:05:45+00:00 https://ainews.lumi-systems.io/2026/05/31/build-a-test-suite-that-grows-with-your-agent-with-dataset-management-in-amazon-bedrock-agentcore/ 2026-07-04T10:05:44+00:00 https://ainews.lumi-systems.io/2026/05/31/designing-ai-resistant-technical-evaluations-2/ 2026-07-04T10:05:43+00:00 https://ainews.lumi-systems.io/2026/05/31/demystifying-evals-for-ai-agents-2/ 2026-07-04T10:05:42+00:00 https://ainews.lumi-systems.io/2026/05/31/effective-harnesses-for-long-running-agents-2/ 2026-07-04T10:05:41+00:00 https://ainews.lumi-systems.io/2026/05/31/introducing-advanced-tool-use-on-the-claude-developer-platform-2/ 2026-07-04T10:05:39+00:00 https://ainews.lumi-systems.io/2026/05/31/kompromittierte-axios-npm-pakete-verbreiten-schadsoftware/ 2026-07-04T10:05:38+00:00 https://ainews.lumi-systems.io/2026/05/31/dns-aid-will-make-ai-agents-easier-to-discover-says-linux-foundation/ 2026-07-04T10:05:37+00:00 https://ainews.lumi-systems.io/2026/05/31/manipuliertes-npm-paket-stiehlt-langlebige-openai-tokens-von-entwicklern/ 2026-07-04T10:05:36+00:00 https://ainews.lumi-systems.io/2026/06/01/itbench-aa-frontier-modelle-verfehlen-50-prozent-marke-bei-enterprise-it-aufgaben/ 2026-07-04T10:05:35+00:00 https://ainews.lumi-systems.io/2026/06/01/flowise-mcp-luecke-ermoeglicht-remote-code-execution-mit-einem-klick/ 2026-07-04T10:05:34+00:00 https://ainews.lumi-systems.io/2026/06/01/audiohijack-versteckte-signale-in-audio-manipulieren-ki-sprachmodelle/ 2026-07-04T10:05:33+00:00 https://ainews.lumi-systems.io/2026/06/01/supply-chain-attacke-miasma-red-hat-npm-pakete-mit-worm-infiziert/ 2026-07-04T10:05:32+00:00 https://ainews.lumi-systems.io/2026/06/02/gpt-5-5-gpt-5-4-und-codex-jetzt-allgemein-verfuegbar-auf-amazon-bedrock/ 2026-07-04T10:05:30+00:00 https://ainews.lumi-systems.io/2026/06/02/mehrheit-agent-systeme-beschleunigen-computernutzungs-agenten-deutlich/ 2026-07-04T10:05:29+00:00 https://ainews.lumi-systems.io/2026/06/02/oauth-code-flow-fuer-sichere-agent-zu-tool-authentifizierung-in-agentcore-gateway/ 2026-07-04T10:05:28+00:00 https://ainews.lumi-systems.io/2026/06/02/malicious-npm-paket-zielt-auf-openai-codex-nutzer-und-deckt-lieferketten-risiken-auf/ 2026-07-04T10:05:27+00:00 https://ainews.lumi-systems.io/2026/06/02/geometric-latent-reasoning-verkuerzt-generierung-in-grossen-sprachmodellen/ 2026-07-04T10:05:26+00:00 https://ainews.lumi-systems.io/2026/06/02/analyse-nlp-forschung-meldet-annotator-details-zu-selektiv/ 2026-07-04T10:05:25+00:00 https://ainews.lumi-systems.io/2026/06/02/project-glasswing-anthropic-erweitert-ki-sicherheitsinitiative-auf-150-neue-partner/ 2026-07-04T10:05:23+00:00 https://ainews.lumi-systems.io/2026/06/02/claude-und-andere-llm-agenten-durch-kombiniertes-policy-und-world-model-training-effizienter/ 2026-07-04T10:05:22+00:00 https://ainews.lumi-systems.io/2026/06/02/harness-1-search-agent-mit-ausgelagertem-zustandsmanagement-durch-rl-trainiert/ 2026-07-04T10:05:20+00:00 https://ainews.lumi-systems.io/2026/06/02/github-plant-agenten-strategie-fuer-codeflut-durch-ki/ 2026-07-04T10:05:19+00:00 https://ainews.lumi-systems.io/2026/06/02/hyperparameter-optimierung-fuer-spezialisierte-modelle-auf-amazon-nova-forge/ 2026-07-04T10:05:18+00:00 https://ainews.lumi-systems.io/2026/06/02/microsoft-bringt-surface-rtx-spark-dev-box-fuer-lokale-ki-entwicklung/ 2026-07-04T10:05:17+00:00 https://ainews.lumi-systems.io/2026/06/02/microsoft-erweitert-windows-11-fuer-lokale-ki-entwicklung/ 2026-07-04T10:05:16+00:00 https://ainews.lumi-systems.io/2026/06/03/omnidreams-nvidia-entwickelt-echtzeit-weltmodell-fuer-autonome-fahrzeugsimulatoren/ 2026-07-04T10:05:14+00:00 https://ainews.lumi-systems.io/2026/06/03/vase-stochastische-kv-cache-eviction-fuer-reasoning-modelle/ 2026-07-04T10:05:13+00:00 https://ainews.lumi-systems.io/2026/06/03/rl-gesteuertes-sampling-fuer-test-time-scaling-bei-large-language-models/ 2026-07-04T10:05:12+00:00 https://ainews.lumi-systems.io/2026/06/03/microsoft-stellt-mai-thinking-1-vor-neues-ki-modell-fuer-coding/ 2026-07-04T10:05:11+00:00 https://ainews.lumi-systems.io/2026/06/03/sicherheitsforscher-veroeffentlicht-0-day-schwachstelle-in-github-und-vscode/ 2026-07-04T10:05:10+00:00 https://ainews.lumi-systems.io/2026/06/03/github-copilot-microsoft-stellt-auf-token-basierte-abrechnung-um/ 2026-07-04T10:05:09+00:00 https://ainews.lumi-systems.io/2026/06/03/microsoft-entwickelt-sicherheitsrahmen-fuer-autonome-ki-agenten/ 2026-07-04T10:05:08+00:00 https://ainews.lumi-systems.io/2026/06/03/context-engineering-die-disziplin-hinter-modernen-ki-systemen/ 2026-07-04T10:05:06+00:00 https://ainews.lumi-systems.io/2026/06/03/linux-foundation-startet-dns-aid-dezentrales-verzeichnis-fuer-ki-agenten/ 2026-07-04T10:05:05+00:00 https://ainews.lumi-systems.io/2026/06/03/nvidia-stellt-physical-ai-agent-skills-fuer-autonome-fahrzeuge-und-robotik-vor/ 2026-07-04T10:05:04+00:00 https://ainews.lumi-systems.io/2026/06/03/praezision-bei-tool-aufrufen-sft-und-dpo-fuer-language-models-auf-sagemaker/ 2026-07-04T10:05:03+00:00 https://ainews.lumi-systems.io/2026/06/03/kvarn-varianzbasierte-kv-cache-quantisierung-reduziert-fehlerakkumulation/ 2026-07-04T10:05:01+00:00 https://ainews.lumi-systems.io/2026/06/04/grail-verbessertes-reinforcement-learning-fuer-mathematisches-reasoning-in-llms/ 2026-07-04T10:05:00+00:00 https://ainews.lumi-systems.io/2026/06/04/meta-agent-challenge-frontier-modelle-scheitern-bei-autonomer-agent-entwicklung/ 2026-07-04T10:04:59+00:00 https://ainews.lumi-systems.io/2026/06/04/stride-trainingsdateneinfluss-in-llms-via-sparse-recovery-nachverfolgen/ 2026-07-04T10:04:58+00:00 https://ainews.lumi-systems.io/2026/06/04/hugging-face-transformers-rce-luecke-in-modellkonfigurationen-umgeht-sicherheitsmassnahmen/ 2026-07-04T10:04:26+00:00 https://ainews.lumi-systems.io/2026/06/04/dar-agentenbasiertes-reasoning-fuer-deontische-logik-und-regelanwendung/ 2026-07-04T10:04:25+00:00 https://ainews.lumi-systems.io/2026/06/04/ironworm-malware-befaellt-36-npm-pakete-in-supply-chain-angriff/ 2026-07-04T10:04:24+00:00 https://ainews.lumi-systems.io/2026/06/05/oprd-representation-distillation-mit-versteckten-zustaenden-schlaegt-output-only-methode/ 2026-07-04T10:04:23+00:00 https://ainews.lumi-systems.io/2026/06/05/dream-exe-video-generierungsmodelle-auf-praktische-robotik-faehigkeiten-testen/ 2026-07-04T10:04:22+00:00 https://ainews.lumi-systems.io/2026/06/05/rubygems-fuehrt-abkuehlungsphase-fuer-paketaktualisierungen-ein/ 2026-07-04T10:04:21+00:00 https://ainews.lumi-systems.io/2026/06/05/wie-reinforcement-learning-umgebungen-trainingsqualitaet-zerstoeren-praktische-loesungen/ 2026-07-04T10:04:19+00:00 https://ainews.lumi-systems.io/2026/06/08/socratic-swe-selbstlernende-ki-agenten-fuer-code-reparaturen/ 2026-07-04T10:04:17+00:00 https://ainews.lumi-systems.io/2026/06/08/ki-agenten-brauchen-observability-statt-marketing-versprechen/ 2026-07-04T10:04:16+00:00 https://ainews.lumi-systems.io/2026/06/08/rise-agentic-search-mit-optimiertem-retrieval-statt-unbegrenzter-corpus-interaktion/ 2026-07-04T10:04:15+00:00 https://ainews.lumi-systems.io/2026/06/09/microsoft-benachrichtigt-kunden-ueber-downloads-infizierter-github-pakete/ 2026-07-04T10:04:14+00:00 https://ainews.lumi-systems.io/2026/06/09/siri-mit-vision-llm-und-eigenem-ai-runtime-auf-apple-hardware/ 2026-07-04T10:04:13+00:00 https://ainews.lumi-systems.io/2026/06/10/adversarische-hacker-fixer-schleifen-schliessen-sicherheitsluecken-in-agent-benchmarks/ 2026-07-04T10:04:10+00:00 https://ainews.lumi-systems.io/2026/06/10/encoder-decoder-architektur-fuer-effiziente-kontext-kompression-in-llms/ 2026-07-04T10:04:08+00:00 https://ainews.lumi-systems.io/2026/06/10/latent-context-language-models-skalierbare-kv-cache-kompression-fuer-lange-kontexte/ 2026-07-04T10:04:07+00:00 https://ainews.lumi-systems.io/2026/06/10/lookahead-sparse-attention-deepseek-v4-reduziert-kv-cache-auf-135-prozent/ 2026-07-04T10:04:06+00:00 https://ainews.lumi-systems.io/2026/06/10/workflow-gym-benchmark-offenbart-grenzen-von-ki-agenten-bei-komplexen-gui-aufgaben/ 2026-07-04T10:04:04+00:00 https://ainews.lumi-systems.io/2026/06/10/pact-effizientere-kommunikation-in-multi-agent-systemen-mit-claude/ 2026-07-04T10:04:03+00:00 https://ainews.lumi-systems.io/2026/06/10/unternehmen-deployen-massenhaft-verwundbare-ki-generierte-software/ 2026-07-04T10:04:02+00:00 https://ainews.lumi-systems.io/2026/06/10/flowtracer-gezielte-reinforcement-learning-in-llms-durch-attention-basierte-informationsflussverfolgung/ 2026-07-04T10:04:01+00:00 https://ainews.lumi-systems.io/2026/06/10/flowtracer-gezieltes-reinforcement-learning-durch-nachverfolgung-von-informationsfluss-in-llms/ 2026-07-04T10:04:00+00:00 https://ainews.lumi-systems.io/2026/06/10/claude-security-ermoeglicht-code-scanning-mit-opus-4-7-auf-unternehmenseigenen-codebasen/ 2026-07-04T10:03:58+00:00 https://ainews.lumi-systems.io/2026/06/10/hybride-llms-verlieren-long-context-faehigkeiten-durch-cot-feintuning/ 2026-07-04T10:03:57+00:00 https://ainews.lumi-systems.io/2026/06/10/symjack-angriff-ueber-symbolische-links-in-ki-coding-agenten/ 2026-07-04T10:03:56+00:00 https://ainews.lumi-systems.io/2026/06/11/frontier-teams-revolutionieren-ai-native-softwareentwicklung-mit-45x-produktivitaetszuwachs/ 2026-07-04T10:03:53+00:00 https://ainews.lumi-systems.io/2026/06/11/github-deaktiviert-automatische-script-ausfuehrung-in-npm-ab-version-12/ 2026-07-04T10:03:52+00:00 https://ainews.lumi-systems.io/2026/06/11/bebop-rejection-sampling-verbessert-multi-token-prediction-in-rl-training/ 2026-07-04T10:03:51+00:00 https://ainews.lumi-systems.io/2026/06/11/claw-swe-bench-benchmark-fuer-ki-agenten-bei-code-aufgaben/ 2026-07-04T10:03:50+00:00 https://ainews.lumi-systems.io/2026/06/11/mixture-of-experts-router-durch-manifold-power-iteration-optimiert/ 2026-07-04T10:03:49+00:00 https://ainews.lumi-systems.io/2026/06/11/github-deaktiviert-npm-installationsskripte-standardmaessig-gegen-supply-chain-angriffe/ 2026-07-04T10:03:48+00:00 https://ainews.lumi-systems.io/2026/06/11/grammar-constrained-decoding-ermoeglicht-jailbreak-von-llms-zur-malware-generierung/ 2026-07-04T10:03:46+00:00 https://ainews.lumi-systems.io/2026/06/11/openai-akquiriert-ona-fuer-erweiterte-cloud-umgebungen/ 2026-07-04T10:03:44+00:00 https://ainews.lumi-systems.io/2026/06/11/openclaw-agenten-durch-praeparierte-eingaben-zur-codeausfuehrung-manipulierbar/ 2026-07-04T10:03:43+00:00 https://ainews.lumi-systems.io/2026/06/12/minimax-sparse-attention-effiziente-long-context-verarbeitung-fuer-milliarden-parameter-modelle/ 2026-07-04T10:03:42+00:00 https://ainews.lumi-systems.io/2026/06/12/langgraph-sql-injection-kette-ermoeglicht-remote-code-execution-in-selbstgehosteten-ki-agenten/ 2026-07-04T10:03:40+00:00 https://ainews.lumi-systems.io/2026/06/15/clinhallu-benchmark-zur-diagnose-von-halluzinationen-in-medizinischen-ki-modellen/ 2026-07-04T10:03:39+00:00 https://ainews.lumi-systems.io/2026/06/15/harnessx-automatisierte-optimierung-von-agent-laufzeitumgebungen/ 2026-07-04T10:03:38+00:00 https://ainews.lumi-systems.io/2026/06/15/sicherheitsmechanismen-von-ki-agenten-als-dos-waffe-missbrauchbar/ 2026-07-04T10:03:37+00:00 https://ainews.lumi-systems.io/2026/06/15/ki-sicherheitssysteme-als-dos-ziel-poisoned-documents-lahmen-guardrails/ 2026-07-04T10:03:36+00:00 https://ainews.lumi-systems.io/2026/06/15/litellm-drei-verkettete-sicherheitsluecken-ermoeglichen-systemuebernahme/ 2026-07-04T10:03:35+00:00 https://ainews.lumi-systems.io/2026/06/16/fastcontext-spezialisierte-agenten-fuer-effiziente-code-repository-erkundung/ 2026-07-04T10:03:33+00:00 https://ainews.lumi-systems.io/2026/06/16/tangram-statische-kv-cache-kompression-fuer-schnelleres-multi-turn-llm-serving/ 2026-07-04T10:03:32+00:00 https://ainews.lumi-systems.io/2026/06/16/ki-in-der-produktentwicklung-vom-isolierten-tool-zum-integrierten-lebenszyklus-system/ 2026-07-04T10:03:31+00:00 https://ainews.lumi-systems.io/2026/06/16/post-training-rezepte-an-der-grenze-von-instructgpt-zu-mopd-spezialistensystemen/ 2026-07-04T10:03:30+00:00 https://ainews.lumi-systems.io/2026/06/16/p-eagle-parallele-spekulation-fuer-schnellere-llm-inferenz-auf-aws-sagemaker/ 2026-07-04T10:03:29+00:00 https://ainews.lumi-systems.io/2026/06/17/amazon-bedrock-invokeguardrailchecks-api-fuer-agentenbasierte-anwendungen/ 2026-07-04T10:03:26+00:00 https://ainews.lumi-systems.io/2026/06/17/loopcoder-v2-zwei-schleifen-als-optimum-fuer-effiziente-modellkomputation-beim-programmieren/ 2026-07-04T10:03:25+00:00 https://ainews.lumi-systems.io/2026/06/17/procua-sft-automatisch-generierte-trainingsdaten-fuer-desktop-agenten/ 2026-07-04T10:03:24+00:00 https://ainews.lumi-systems.io/2026/06/17/ki-integration-scheitert-an-veralteten-it-architekturen/ 2026-07-04T10:03:23+00:00 https://ainews.lumi-systems.io/2026/06/17/144-mastra-npm-pakete-ueber-gekapertes-contributor-konto-kompromittiert/ 2026-07-04T10:03:22+00:00 https://ainews.lumi-systems.io/2026/06/17/bucket-squatting-luecke-in-vertex-ai-sdk-ermoeglichte-modell-hijacking/ 2026-07-04T10:03:21+00:00 https://ainews.lumi-systems.io/2026/06/17/estland-plant-digitale-identitaeten-fuer-ki-agenten-mit-definierten-rechten/ 2026-07-04T10:03:19+00:00 https://ainews.lumi-systems.io/2026/06/17/kostenlose-codeproduktion-dreht-oekonomie-der-softwareentwicklung-um/ 2026-07-04T10:03:18+00:00 https://ainews.lumi-systems.io/2026/06/18/glm-5-2-chinesisches-open-weights-modell-mit-753-milliarden-parametern/ 2026-07-04T10:03:17+00:00 https://ainews.lumi-systems.io/2026/06/19/efficientrollout-selbstspekulative-dekodierung-fuer-schnellere-rl-rollouts/ 2026-07-04T10:03:16+00:00 https://ainews.lumi-systems.io/2026/06/19/reves-iteratives-training-fuer-effizientere-test-time-skalierung-bei-llms/ 2026-07-04T10:03:15+00:00 https://ainews.lumi-systems.io/2026/06/19/vertex-ai-sdk-schwachstelle-ermoeglichte-modell-hijacking-auf-google-cloud/ 2026-07-04T10:03:14+00:00 https://ainews.lumi-systems.io/2026/06/19/microsoft-warnt-vor-remote-code-execution-durch-web-aktivierte-ki-agenten/ 2026-07-04T10:03:12+00:00 https://ainews.lumi-systems.io/2026/06/19/uniform-fp4-neue-4-bit-trainingsmethode-fuer-llms-reduziert-systematische-fehler/ 2026-07-04T10:03:11+00:00 https://ainews.lumi-systems.io/2026/06/19/ki-agenten-als-neue-identitaet-governance-luecken-im-unternehmen/ 2026-07-04T10:03:10+00:00 https://ainews.lumi-systems.io/2026/06/19/autojack-ai-browsing-agenten-fuer-code-ausfuehrung-auf-host-systemen-missbraucht/ 2026-07-04T10:03:09+00:00 https://ainews.lumi-systems.io/2026/06/21/auggie-cli-ki-coding-agent-im-terminal-workflow/ 2026-07-04T10:03:07+00:00 https://ainews.lumi-systems.io/2026/06/22/harness-engineering-kontrollmechanismen-fuer-zuverlaessige-ki-agenten/ 2026-07-04T10:03:05+00:00 https://ainews.lumi-systems.io/2026/06/22/aws-continuum-automatisiert-schwachstellen-handling-mit-ki-gestuetzter-validierung/ 2026-07-04T10:03:04+00:00 https://ainews.lumi-systems.io/2026/06/22/microsoft-behebt-schwachstelle-in-autogen-studio-fuer-code-ausfuehrung/ 2026-07-04T10:03:02+00:00 https://ainews.lumi-systems.io/2026/06/23/github-actions-sicherung-gegen-pull-request-target-missbrauch-ab-version-7/ 2026-07-04T10:03:01+00:00 https://ainews.lumi-systems.io/2026/06/23/evoembedding-kontextabhaengige-embeddings-fuer-lange-sequenzen/ 2026-07-04T10:03:00+00:00 https://ainews.lumi-systems.io/2026/06/23/github-sperrt-actions-checkout-gegen-pwn-request-angriffe/ 2026-07-04T10:02:58+00:00 https://ainews.lumi-systems.io/2026/06/23/parallelkernelbench-frontier-llms-scheitern-noch-an-schnellen-multi-gpu-kerneln/ 2026-07-04T10:02:57+00:00 https://ainews.lumi-systems.io/2026/06/24/naturebench-wie-weit-coding-agenten-in-wissenschaftlichen-aufgaben-wirklich-kommen/ 2026-07-04T10:02:56+00:00 https://ainews.lumi-systems.io/2026/06/24/edv-framework-reduziert-fehlerakkumulation-in-selbstlernenden-llm-agenten/ 2026-07-04T10:02:54+00:00 https://ainews.lumi-systems.io/2026/06/24/ci-cd-schwachstelle-cordyceps-gefaehrdet-github-repositories-ueber-supply-chain-angriffe/ 2026-07-04T10:02:53+00:00 https://ainews.lumi-systems.io/2026/06/25/github-blockiert-automatische-code-checkouts-aus-geforkten-pull-requests/ 2026-07-04T10:02:52+00:00 https://ainews.lumi-systems.io/2026/06/26/tool-calling-ausfaelle-unter-schema-constraints-in-open-weight-llms/ 2026-07-04T10:02:49+00:00 https://ainews.lumi-systems.io/2026/06/26/sprachkompression-in-llms-output-optimierung-spart-kosten-input-reduktion-erhoeht-sie/ 2026-07-04T10:02:48+00:00 https://ainews.lumi-systems.io/2026/06/26/viq-diskrete-visuelle-darstellungen-auf-beliebiger-aufloesung/ 2026-07-04T10:02:47+00:00 https://ainews.lumi-systems.io/2026/06/26/openbiorq-benchmark-fuer-agentenbasierte-biomedizinische-forschungsfragen/ 2026-07-04T10:02:45+00:00 https://ainews.lumi-systems.io/2026/06/26/openbiorq-benchmark-fuer-agentic-ki-modelle-in-biomedizinischen-forschungsfragen/ 2026-07-04T10:02:43+00:00 https://ainews.lumi-systems.io/2026/06/26/ki-tools-in-der-softwareentwicklung-sicherheitsluecken-trotz-compliance-bemuehungen/ 2026-07-04T10:02:42+00:00 https://ainews.lumi-systems.io/2026/06/26/gauntletbench-neue-messlatte-zeigt-grenzen-von-ki-agenten-auf/ 2026-07-04T10:02:41+00:00 https://ainews.lumi-systems.io/2026/06/26/stripe-skaliert-compliance-mit-ki-agenten-26-prozent-schneller-bei-voller-kontrolle/ 2026-07-04T10:02:39+00:00 https://ainews.lumi-systems.io/2026/06/26/infokv-entropie-basierte-kv-cache-kompression-fuer-lange-reasoning-sequenzen/ 2026-07-04T10:02:38+00:00 https://ainews.lumi-systems.io/2026/06/27/github-repository-trick-taeuscht-ki-agenten-in-ausfuehrung-von-schadsoftware/ 2026-07-04T10:02:37+00:00 https://ainews.lumi-systems.io/2026/06/28/25-prozent-deutscher-unternehmen-haben-keine-kontrolle-ueber-ki-entwicklungstools/ 2026-07-04T10:02:35+00:00 https://ainews.lumi-systems.io/2026/06/28/lokale-sprachmodelle-in-die-anwendung-integrieren-von-ollama-zum-produktiven-code/ 2026-07-04T10:02:34+00:00 https://ainews.lumi-systems.io/2026/06/29/ki-generierte-codeprojekte-zeigen-systematische-sicherheitsmaengel/ 2026-07-04T10:02:33+00:00 https://ainews.lumi-systems.io/2026/06/29/ornith-1-0-open-source-modell-fuer-agentengesteuerte-softwareentwicklung/ 2026-07-04T10:02:31+00:00 https://ainews.lumi-systems.io/2026/06/29/multi-tenant-llm-analytik-mit-row-level-security-par-technologys-architektur-auf-aws/ 2026-07-04T10:02:30+00:00 https://ainews.lumi-systems.io/2026/06/30/vision-ai-agenten-synthetic-data-und-fine-tuning-fuer-hoehere-genauigkeit/ 2026-07-04T10:02:28+00:00 https://ainews.lumi-systems.io/2026/06/30/guardfall-ki-coding-agenten-anfaellig-fuer-klassische-shell-injection-tricks/ 2026-07-04T10:02:27+00:00 https://ainews.lumi-systems.io/2026/06/30/asynchrone-pipeline-parallelisierung-fuer-llm-vortraining-bei-gradient-verzoegerung-praktikabel/ 2026-07-04T10:02:26+00:00 https://ainews.lumi-systems.io/2026/06/30/35b-agentenmodell-erreicht-leistung-von-billionen-parameter-systemen-durch-horizont-skalierung/ 2026-07-04T10:02:24+00:00 https://ainews.lumi-systems.io/2026/06/30/google-stellt-evaluierungs-framework-fuer-code-agenten-vor/ 2026-07-04T10:02:22+00:00 https://ainews.lumi-systems.io/2026/06/30/microsoft-warnt-vor-vergifteten-mcp-tool-beschreibungen-als-datenabzugsvektor/ 2026-07-04T10:02:21+00:00 https://ainews.lumi-systems.io/2026/07/01/gefaelschte-bug-reports-kapern-ki-coding-agenten-im-grossen-stil/ 2026-07-04T10:02:19+00:00 https://ainews.lumi-systems.io/2026/07/01/lumos-semantische-os-schicht-fuer-zugaenglichkeits-gestuetzte-ki-agenten/ 2026-07-04T10:02:18+00:00 https://ainews.lumi-systems.io/2026/07/01/enterprise-ki-erfordert-portfolio-ansatz-statt-single-model-strategie/ 2026-07-04T10:02:17+00:00