https://newskeryx.com/brief/yoshua-bengio-urges-safety-focused-people-to-leave-frontier-ai-cNewskeryxen2026-10-20Yoshua Bengio urges safety-focused people to leave frontier AI companies https://newskeryx.com/brief/yelp-adds-ai-agent-that-calls-restaurants-to-check-waits-and-joiNewskeryxen2026-10-20Yelp adds AI agent that calls restaurants to check waits and join waitlists https://newskeryx.com/brief/us-orders-civilian-agencies-to-remove-deepseek-ai-tools-over-secNewskeryxen2026-10-20US orders civilian agencies to remove DeepSeek AI tools over security concerns https://newskeryx.com/brief/several-smaller-ai-model-releases-land-including-a-fast-ocr-modeNewskeryxen2026-10-20Several smaller AI model releases land, including a fast OCR model https://newskeryx.com/brief/princeton-researchers-publish-a-review-of-how-ai-agents-are-testNewskeryxen2026-10-20Princeton researchers publish a review of how AI agents are tested https://newskeryx.com/brief/opencode-reports-48-hour-agent-swarm-attack-some-ai-models-refusNewskeryxen2026-10-20OpenCode reports 48-hour agent-swarm attack; some AI models refused to help analyse it https://newskeryx.com/brief/open-source-tool-released-for-tracing-and-debugging-ai-apps-andNewskeryxen2026-10-20Open-source tool released for tracing and debugging AI apps and agents https://newskeryx.com/brief/newsletter-argues-ai-adoption-in-other-sectors-will-follow-techNewskeryxen2026-10-20Newsletter argues AI adoption in other sectors will follow tech's path https://newskeryx.com/brief/minimax-reveals-anonymous-space-bunny-model-was-m3-1-flash-releaNewskeryxen2026-10-20MiniMax reveals anonymous Space Bunny model was M3.1 Flash, releases it cheaper https://newskeryx.com/brief/meta-releases-muse-models-keeps-its-strongest-one-subscription-oNewskeryxen2026-10-20Meta releases Muse models, keeps its strongest one subscription-only https://newskeryx.com/brief/inception-launches-mercury-3-a-fast-diffusion-based-language-modNewskeryxen2026-10-20Inception launches Mercury 3, a fast diffusion-based language model https://newskeryx.com/brief/guide-shows-splitting-one-language-model-across-two-gpu-serversNewskeryxen2026-10-20Guide shows splitting one language model across two GPU servers on AWS https://newskeryx.com/brief/fal-and-bfl-release-flux-3-max-a-video-generating-modelNewskeryxen2026-10-20fal and BFL release FLUX 3 Max, a video-generating model https://newskeryx.com/brief/essay-argues-ai-coding-agents-are-eroding-programming-as-a-craftNewskeryxen2026-10-20Essay argues AI coding agents are eroding programming as a craft https://newskeryx.com/brief/democrats-prepare-to-fight-over-ai-regulationNewskeryxen2026-10-20Democrats prepare to fight over AI regulation https://newskeryx.com/brief/deepseek-releases-updated-flash-model-claimed-top-open-weight-vaNewskeryxen2026-10-20DeepSeek releases updated Flash model, claimed top open-weight value https://newskeryx.com/brief/browser-agents-can-now-pay-for-web-resources-under-enforced-budgNewskeryxen2026-10-20Browser agents can now pay for web resources under enforced budgets https://newskeryx.com/brief/tenstorrent-raises-1-5-billion-to-challenge-nvidia-in-ai-chipsNewskeryxen2026-10-19Tenstorrent raises $1.5 billion to challenge Nvidia in AI chips https://newskeryx.com/brief/simon-willison-releases-llm-echo-0-5-a-testing-plugin-for-his-llNewskeryxen2026-10-19Simon Willison releases llm-echo 0.5, a testing plugin for his LLM tool https://newskeryx.com/brief/openai-dismisses-three-safety-staff-citing-disputed-reasonsNewskeryxen2026-10-19OpenAI dismisses three safety staff, citing disputed reasons https://newskeryx.com/brief/epoch-tests-six-frontier-ai-models-on-real-research-tasksNewskeryxen2026-10-19Epoch grades six frontier AI models on real research tasks https://newskeryx.com/brief/anthropic-s-jack-clark-says-self-improving-ai-may-arrive-by-2027Newskeryxen2026-10-19Anthropic's Jack Clark says self-improving AI may arrive by 2027 https://newskeryx.com/brief/anthropic-reports-fewer-junior-hires-and-more-time-spent-in-meetNewskeryxen2026-10-19Anthropic reports fewer junior hires and more time spent in meetings https://newskeryx.com/brief/anthropic-proposes-ai-safety-bill-triggered-by-50-deaths-or-1-biNewskeryxen2026-10-19Anthropic proposes AI safety bill triggered by 50 deaths or $1 billion damages https://newskeryx.com/brief/anthropic-pledges-5-of-its-shares-to-a-public-benefit-fundNewskeryxen2026-10-19Anthropic pledges 5% of its shares to a public benefit fund https://newskeryx.com/brief/states-sue-openai-seeking-to-bar-self-improving-ai-modelsNewskeryxen2026-10-18US states ask court to bar OpenAI from building self-improving AI https://newskeryx.com/brief/stanford-student-uses-ai-model-to-design-viruses-in-2025Newskeryxen2026-10-18Stanford student uses AI model to design viruses in 2025 https://newskeryx.com/brief/rj-scaringe-says-robots-are-advancing-faster-than-expectedNewskeryxen2026-10-18Rivian CEO says robotics is advancing faster than people expect https://newskeryx.com/brief/openai-s-gpt-6-astra-report-flags-weaker-view-into-model-reasoniNewskeryxen2026-10-18OpenAI's GPT-6 Astra shows less of its written reasoning https://newskeryx.com/brief/openai-and-anthropic-discuss-ai-agent-autonomy-openai-fires-threNewskeryxen2026-10-18OpenAI and Anthropic discuss AI agent autonomy; OpenAI fires three safety staff https://newskeryx.com/brief/meta-s-top-muse-wonder-model-is-subscription-only-not-availableNewskeryxen2026-10-18Meta releases Muse Wonder models, strongest version limited to subscribers https://newskeryx.com/brief/kevin-roose-questions-whether-ai-industry-can-be-trusted-with-poNewskeryxen2026-10-18Kevin Roose discusses AI industry trust, safety and rivalries on podcast https://newskeryx.com/brief/jpmorgan-s-ceo-urges-regulators-to-prepare-for-banking-ai-agentsNewskeryxen2026-10-18JPMorgan's CEO urges regulators to prepare for banking AI agents https://newskeryx.com/brief/free-open-source-course-teaches-building-coding-agents-condensedNewskeryxen2026-10-18Free open-source course teaches building coding agents, condensed into 90-minute webinar https://newskeryx.com/brief/anthropic-s-jack-clark-forecasts-recursive-ai-self-improvement-bNewskeryxen2026-10-18Anthropic's Jack Clark Forecasts Recursive AI Self-Improvement by 2027