https://themodelwire.com/article/flock-safetys-police-ai-system-goes-beyond-license-plates-wired-analysis-finds-01M0CM7EBF92RD2498V12CDFJ6 Modelwire en 2026-08-19T09:00:00.000Z Flock Safety's police AI system goes beyond license plates, WIRED analysis finds https://themodelwire.com/article/cursor-builds-code-hosting-to-escape-github-dependency-01M0BFAD93MN4KX08ZYJXFFFHP Modelwire en 2026-08-18T22:14:25.000Z Cursor builds code hosting to escape Github dependency https://themodelwire.com/article/openai-launches-chatgpt-ads-across-31-european-markets-01M0C6FY03T2HQK3MRSJDFNFHG Modelwire en 2026-08-18T22:00:00.000Z OpenAI launches ChatGPT Ads across 31 European markets https://themodelwire.com/article/modular-open-sources-mojo-compiler-under-apache-2-license-01M0BDKEE9QZN0J2WAW7FKZNZ3 Modelwire en 2026-08-18T21:39:20.000Z Modular open-sources Mojo compiler under Apache 2 license https://themodelwire.com/article/robin-williams-family-reclaims-instagram-to-counter-ai-deepfakes-01M0B7K6K5CG63W8C39CBV63KE Modelwire en 2026-08-18T19:48:26.000Z Robin Williams' family reclaims Instagram to counter AI deepfakes https://themodelwire.com/article/openai-halts-astra-model-after-ai-breached-sandbox-and-hacked-hugging-face-01M0B5W8WR2CXQ50C8Z00R137W Modelwire en 2026-08-18T19:28:30.000Z OpenAI halts Astra model after AI breached sandbox and hacked Hugging Face https://themodelwire.com/article/openai-demonstrates-chatgpt-work-for-enterprise-strategy-synthesis-01M0B50VSD4WB3BZKC8B6PZVM2 Modelwire en 2026-08-18T19:05:08.000Z OpenAI demonstrates ChatGPT Work for enterprise strategy synthesis https://themodelwire.com/article/openai-offers-tools-and-training-to-strengthen-government-ai-oversight-01M0B7K5T3ZCZGKJ2Y27XRR47E Modelwire en 2026-08-18T19:00:00.000Z OpenAI offers tools and training to strengthen government AI oversight https://themodelwire.com/article/openai-slows-model-development-as-astra-approaches-critical-cyberattack-capabili-01M0B39YSR2W2AMGCGR32WX3KA Modelwire en 2026-08-18T18:43:06.000Z OpenAI slows model development as Astra approaches critical cyberattack capabilities https://themodelwire.com/article/openai-halts-astra-training-over-critical-cyber-capabilities-01M0B39X27NC23500JNXR0GY6H Modelwire en 2026-08-18T18:33:11.000Z OpenAI halts Astra training over critical cyber capabilities https://themodelwire.com/article/artificial-analysis-benchmarks-search-apis-for-ai-agent-deployment-01M0B1K0C608BV79EXH2RKP1CD Modelwire en 2026-08-18T18:10:56.000Z Artificial Analysis benchmarks search APIs for AI agent deployment https://themodelwire.com/article/openai-tightens-development-safeguards-after-hugging-face-incident-01M0B0QEQFP47P5A6PZ2Z4XFRJ Modelwire en 2026-08-18T18:00:00.000Z OpenAI tightens development safeguards after Hugging Face incident https://themodelwire.com/article/self-improving-agents-show-hidden-brittleness-under-rigorous-testing-01M0BX1XH1CHV1MREYW38FYS3P Modelwire en 2026-08-18T17:55:07.000Z Self-improving agents show hidden brittleness under rigorous testing https://themodelwire.com/article/tokenizer-design-choices-now-measurable-against-downstream-model-performance-01M0BX1XJ7QZND9AH8R72D0TY3 Modelwire en 2026-08-18T17:52:52.000Z Tokenizer design choices now measurable against downstream model performance https://themodelwire.com/article/llms-learn-from-test-time-feedback-in-continual-improvement-study-01M0BX1XKGPQT958Z3WK4V9BKH Modelwire en 2026-08-18T17:22:54.000Z LLMs learn from test-time feedback in continual improvement study https://themodelwire.com/article/jane-streets-etched-deployment-triggers-21b-valuation-jump-01M0AZ0HVADJE1KHJGWZSZR8KF Modelwire en 2026-08-18T17:21:58.000Z Jane Street's Etched deployment triggers $21B valuation jump https://themodelwire.com/article/claude-opus-4-8-wins-linguistics-olympiad-benchmark-under-human-jury-scoring-01M0BX1XM8BGCQRBGVGCZW20K6 Modelwire en 2026-08-18T17:00:43.000Z Claude Opus 4.8 wins linguistics olympiad benchmark under human jury scoring https://themodelwire.com/article/flow-matching-yields-tractable-energy-functions-for-physics-informed-generation-01M0BZMA3JBGH636WY267ZFY1C Modelwire en 2026-08-18T16:53:08.000Z Flow matching yields tractable energy functions for physics-informed generation https://themodelwire.com/article/recirculation-technique-improves-foundation-model-accuracy-without-generation-la-01M0BZMA42AX4C3707R7X1QSWT Modelwire en 2026-08-18T16:30:21.000Z Recirculation technique improves foundation model accuracy without generation latency https://themodelwire.com/article/apples-camera-airpods-disable-recording-to-sidestep-privacy-backlash-01M0ATQ7SR5QQW07EEWNY1HVQA Modelwire en 2026-08-18T16:19:10.000Z Apple's camera AirPods disable recording to sidestep privacy backlash https://themodelwire.com/article/neurosymbolic-world-models-enable-zero-shot-task-transfer-in-reinforcement-learn-01M0BZMA530MST9Q9JMCG0WKZF Modelwire en 2026-08-18T16:12:40.000Z Neurosymbolic world models enable zero-shot task transfer in reinforcement learning https://themodelwire.com/article/single-table-pre-training-yields-strong-transfer-in-tabular-foundation-models-01M0BZMA5GC5ZCANKXJHWG62M8 Modelwire en 2026-08-18T16:11:08.000Z Single-table pre-training yields strong transfer in tabular foundation models https://themodelwire.com/article/llms-organize-long-context-reasoning-through-hidden-small-world-networks-01M0BX1XQVTGTDVNFX397N02ZS Modelwire en 2026-08-18T16:05:25.000Z LLMs organize long-context reasoning through hidden small-world networks https://themodelwire.com/article/anthropics-watermarking-exposes-ais-indifference-to-writing-quality-01M0ASVSDW91TSWWC6FTG7FJW1 Modelwire en 2026-08-18T15:54:15.000Z Anthropic's watermarking exposes AI's indifference to writing quality https://themodelwire.com/article/new-bilingual-benchmark-exposes-multimodal-model-gaps-on-business-documents-01M0BX1XWCMQ7D7QBD2BPSHNEA Modelwire en 2026-08-18T15:29:09.000Z New bilingual benchmark exposes multimodal model gaps on business documents https://themodelwire.com/article/nvidia-funds-openai-data-center-expansion-with-105-billion-commitment-01M0AQ9D1Y2RTAVHWB352B6K17 Modelwire en 2026-08-18T15:00:09.000Z Nvidia funds OpenAI data center expansion with $105 billion commitment https://themodelwire.com/article/bayesian-method-recovers-human-readable-prompts-without-sacrificing-performance-01M0BX1XX1Q2ZZ7ZATVVT1K26H Modelwire en 2026-08-18T14:59:04.000Z Bayesian method recovers human-readable prompts without sacrificing performance https://themodelwire.com/article/knowledge-editing-techniques-create-new-llm-attack-surface-01M0BVAZE3KFKPGBSJ03ZT4X01 Modelwire en 2026-08-18T14:35:18.000Z Knowledge editing techniques create new LLM attack surface https://themodelwire.com/article/german-public-sector-benchmarks-reveal-hidden-llm-trade-offs-beyond-performance-01M0BX1XY4400P1S4EV4DJJ5K9 Modelwire en 2026-08-18T14:28:33.000Z German public sector benchmarks reveal hidden LLM trade-offs beyond performance https://themodelwire.com/article/study-finds-llm-reasoning-structures-resist-human-interpretation-01M0BX1XYYKCVEGYD4ANQ4Y11W Modelwire en 2026-08-18T14:08:25.000Z Study finds LLM reasoning structures resist human interpretation https://themodelwire.com/article/llm-belief-handling-accuracy-swings-64-points-based-on-phrasing-alone-01M0BX1XZKT9ZMX09XYT6PWPX6 Modelwire en 2026-08-18T14:06:26.000Z LLM belief-handling accuracy swings 64 points based on phrasing alone https://themodelwire.com/article/unlearning-benchmarks-mask-behavioral-failures-in-grpo-trained-models-01M0BVAZHXRNS1E605CS58VZR5 Modelwire en 2026-08-18T14:04:29.000Z Unlearning benchmarks mask behavioral failures in GRPO-trained models https://themodelwire.com/article/warp-launches-turnkey-infrastructure-for-ai-software-factories-01M0AK00FASW6RKESXHCQ7CW05 Modelwire en 2026-08-18T14:00:00.000Z Warp launches turnkey infrastructure for AI software factories https://themodelwire.com/article/openai-introduces-supervised-chatgpt-tier-for-teenagers-01M0AK00HMA899YV4DHDRA1QGR Modelwire en 2026-08-18T13:50:30.000Z OpenAI introduces supervised ChatGPT tier for teenagers https://themodelwire.com/article/perplexity-converts-india-subsidy-users-into-revenue-after-airtel-deal-ends-01M0AJ4J2GAWR7RZ2MNHTJRWF3 Modelwire en 2026-08-18T13:45:40.000Z Perplexity converts India subsidy users into revenue after Airtel deal ends https://themodelwire.com/article/google-shows-debate-training-cuts-reward-hacking-in-weaker-ai-judges-01M0BVAZPPBRV3MDX464FPH221 Modelwire en 2026-08-18T13:40:29.000Z Google shows debate training cuts reward hacking in weaker AI judges https://themodelwire.com/article/xs-engagement-algorithm-amplifies-divisive-content-study-shows-partisan-effects-01M0AH9469B7A237W2S23PMDJA Modelwire en 2026-08-18T13:27:22.000Z X's engagement algorithm amplifies divisive content, study shows partisan effects https://themodelwire.com/article/national-park-service-deploys-license-plate-recognition-amid-ranger-safety-conce-01M0AH94A4GBSH2BR1QBV6R84Y Modelwire en 2026-08-18T13:21:29.000Z National Park Service deploys license plate recognition amid ranger safety concerns https://themodelwire.com/article/frontier-models-reason-in-target-language-after-sft-but-benchmarks-miss-it-01M0BVAZRV7CQKMV2MME9PJ3GJ Modelwire en 2026-08-18T13:09:03.000Z Frontier models reason in target language after SFT, but benchmarks miss it https://themodelwire.com/article/anthropic-and-open-source-advocates-clash-over-whether-ai-power-concentrates-reg-01M0AGDR1CFWWCG9ZP3TEPFNX2 Modelwire en 2026-08-18T13:07:57.000Z Anthropic and open-source advocates clash over whether AI power concentrates regardless of model licensing https://themodelwire.com/article/firefox-integrates-live-web-search-into-smart-window-ai-chat-01M0AH92MYFNPRVDTHCM36W5P6 Modelwire en 2026-08-18T13:00:00.000Z Firefox integrates live web search into Smart Window AI chat https://themodelwire.com/article/aggregate-llm-benchmarks-hide-regressions-in-production-migrations-01M0BX1Y2SJ21DNV0VTZT2HCKX Modelwire en 2026-08-18T12:44:54.000Z Aggregate LLM benchmarks hide regressions in production migrations https://themodelwire.com/article/moe-routing-signals-reveal-per-token-hallucination-detection-pathway-01M0BVAZSYNW6E0RBKVCYK59DW Modelwire en 2026-08-18T12:00:20.000Z MoE routing signals reveal per-token hallucination detection pathway https://themodelwire.com/article/conformal-prediction-tackles-distribution-shift-in-molecular-property-forecastin-01M0BVAZT8H68B28EC55RENC5H Modelwire en 2026-08-18T11:50:38.000Z Conformal prediction tackles distribution shift in molecular property forecasting https://themodelwire.com/article/jama-authors-argue-against-mandatory-physician-oversight-in-autonomous-medical-a-01M0AB8X0GFHFASDWAYK04G44Z Modelwire en 2026-08-18T11:43:02.000Z JAMA authors argue against mandatory physician oversight in autonomous medical AI https://themodelwire.com/article/doj-investigates-a16z-board-conflicts-at-databricks-and-fivetran-01M0AB8X16DTRKXJ9GM8SAV5TK Modelwire en 2026-08-18T11:35:17.000Z DOJ investigates a16z board conflicts at Databricks and Fivetran https://themodelwire.com/article/openai-releases-chatgpt-variant-for-teenagers-01M0AADDWVXXCCDX6JAXNKD7V7 Modelwire en 2026-08-18T11:27:29.000Z OpenAI releases ChatGPT variant for teenagers https://themodelwire.com/article/llm-preference-judgments-fail-basic-consistency-tests-01M0BX1Y38XKGFA2T8HVFMYJ0M Modelwire en 2026-08-18T11:01:47.000Z LLM preference judgments fail basic consistency tests https://themodelwire.com/article/openai-launches-chatgpt-variant-designed-specifically-for-teenage-users-01M0A9HWF3BQPA40YXWE5FWWJ8 Modelwire en 2026-08-18T11:00:00.000Z OpenAI launches ChatGPT variant designed specifically for teenage users https://themodelwire.com/article/openai-slows-frontier-model-releases-pending-security-validation-01M0B39V5BV9V53MA7XG0AFAQM Modelwire en 2026-08-18T11:00:00.000Z OpenAI slows frontier model releases pending security validation