エピソード

  • NVIDIA’s Hugging Face Deal Puts AI’s Shared Infrastructure in Play

    NVIDIA’s Hugging Face Deal Puts AI’s Shared Infrastructure in Play
    2026/09/03
    NVIDIA agrees to acquire Hugging Face for $12.93 billion, bringing a major open-model platform closer to the world’s leading AI-chip supplier. Resident expert Maya Chen examines what genuine platform neutrality would require. Jesse also covers OpenAI’s billion-dollar cyber-defence programme, Australia’s government AI-assurance framework, the GitSpawn coding-agent vulnerability, World Labs’ Atlas model, urgent SonicWall patches, overlapping AI-service outages and Google’s WeatherNext 3 rollout. Full transcript: https://www.owenonthenet.com/podcast/2026-09-04/ Sources: https://blogs.nvidia.com/blog/nvidia-to-acquire-hugging-face https://apnews.com/article/d96d50e037a2ade479dcdf81cdf2afcf https://openai.com/index/daybreak-for-frontline-defenders https://www.axios.com/2026/09/03/openai-critical-infrastructure-cyber-ai-models https://www.finance.gov.au/government/public-data/data-and-digital-ministers-meeting/national-framework-assurance-artificial-intelligence-government/statement-data-and-digital-ministers https://www.finance.gov.au/government/public-data/data-and-digital-ministers-meeting/national-framework-assurance-artificial-intelligence-government/implementing-australias-ai-ethics-principles-government https://www.manifold.security/blog/ai-coding-agents-git-hijack https://blog.google/innovation-and-ai/models-and-research/google-deepmind/introducing-weathernext-3 https://www.worldlabs.ai/blog/atlas https://www.sonicwall.com/support/notices/product-notice-sma-1000-series-affected-by-multiple-vulnerabilities-snwlid-2026-0016/kA1VN000002AXmQ0AW https://status.openai.com/incidents/2rm6gqeh https://status.claude.com/ https://status.x.ai/grok-com/INC25664c15 https://www.datacenterdynamics.com/en/news/chatgpt-claude-and-grok-hit-by-simultaneous-outages
    続きを読む 一部表示
    22 分
  • Gemini’s two-tier release puts capability behind different gates

    Gemini’s two-tier release puts capability behind different gates
    2026/09/03
    Google opens Gemini 3.8 Flash broadly while reserving a more permissive cyber model for approved defenders. Also: Anthropic’s reference agents expose the unfinished work behind AI commerce, Gemini learns to inspect selected moments in video, Qwen combines language with 3D driving perception and planning, the US Justice Department backs fair use for model training, Cisco patches a critical Nexus switch flaw, Broadcom reports surging AI infrastructure revenue, and the RBA focuses on tokenised wholesale settlement rather than a retail digital currency. Full transcript: https://www.owenonthenet.com/podcast/2026-09-03/ Sources: https://blog.google/innovation-and-ai/models-and-research/gemini-models/3-8-flash-and-3-8-flash-cyber https://deepmind.google/models/model-cards/gemini-3-8-flash https://arstechnica.com/ai/2026/09/google-releases-gemini-3-8-flash-its-third-flash-model-in-six-weeks https://claude.com/blog/claude-for-commerce-agents https://github.com/anthropics/commerce-agents https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-agentic-video-in-gemini https://ai.google.dev/gemini-api/docs/video-understanding https://qwen.ai/blog?id=qwen-drive-1.0 https://arxiv.org/abs/2609.00111 https://storage.courtlistener.com/recap/gov.uscourts.nysd.640396/gov.uscourts.nysd.640396.1682.0.pdf https://apnews.com/article/justice-department-new-york-times-openai-copyright-dbb22e8e02c660ee5a8644915dec39a4 https://sec.cloudapps.cisco.com/security/center/content/CiscoSecurityAdvisory/cisco-sa-n9k-s1-rce-EH8dEtr https://investors.broadcom.com/news-releases/news-release-details/broadcom-inc-announces-third-quarter-fiscal-year-2026-financial https://www.marketscreener.com/news/broadcom-forecasts-quarterly-revenue-below-estimates-ce7858d3db80ff24 https://www.rba.gov.au/media-releases/2026/mr-26-24.html https://www.rba.gov.au/payments-and-infrastructure/tokenised-money/the-role-of-rits-in-supporting-settlement-in-a-tokenised-ecosystem https://www.rba.gov.au/payments-and-infrastructure/tokenised-money/pdf/retail-cbdc-position-paper-sep-2026.pdf
    続きを読む 一部表示
    21 分
  • Astra Crosses the Cyber Threshold as AI Control Moves Into the Product

    Astra Crosses the Cyber Threshold as AI Control Moves Into the Product
    2026/09/03
    OpenAI says its forthcoming Astra model has crossed the company’s Critical cybersecurity threshold after finding two previously unknown vulnerabilities and using them in an exploit chain. We examine why restricted access and runtime intervention are becoming part of frontier-model architecture. Also covered: Australia’s proposed privacy expansion, ChatGPT’s read-only Epic integration, Europe’s €387.8 million LUMI-AI supercomputer, Microsoft’s runtime controls for agents, and BigQuery Graph reaching general availability. In What Changes for You, Claude Fable 5.1 becomes available to developers while NSW limits take-home HSC assessment as generative AI complicates authorship. Full transcript: https://www.owenonthenet.com/podcast/2026-09-02/ Sources: https://openai.com/index/path-to-astra https://www.axios.com/2026/09/01/openai-astras-cyber-critical https://www.anthropic.com/claude-fable-and-mythos-5-1 https://www.axios.com/2026/09/01/anthropic-releases-new-models-cost-structures-and-safeguards https://ministers.ag.gov.au/media-centre/modernising-australias-privacy-laws-digital-age-31-08-2026 https://consultations.ag.gov.au/rights-and-protections/privacy-reform https://openai.com/index/chatgpt-connects-health-records-and-healthcare-sources https://help.openai.com/en/articles/20001490-using-the-epic-plugin-with-chatgpt-and-codex https://www.eurohpc-ju.europa.eu/eurohpc-ju-signs-contract-deploy-lumi-ai-supercomputer-2026-08-31_en https://newsroom.amd.com/news/amd-instinct-gpus-epyc-cpus-power-lumi-ai-supercomputer https://www.nsw.gov.au/ministerial-releases/new-rules-ai-education https://blogs.microsoft.com/on-the-issues/2026/09/01/responsible-ai-in-2026-how-we-are-adapting-for-whats-ahead https://cloud.google.com/blog/products/data-analytics/bigquery-graph-connecting-data-and-ai-at-scale
    続きを読む 一部表示
    20 分
  • Australia’s AI gap is organisational, not technological

    Australia’s AI gap is organisational, not technological
    2026/09/01
    Treasury finds that Australian businesses are experimenting widely with AI but rarely changing work deeply enough to unlock substantial productivity gains. OpenAI expands advertising around ChatGPT conversations, NVIDIA invests in MediaTek as NVLink reaches custom accelerators, and Broadcom brings private AI infrastructure and agent controls into VMware. We also examine urgent ServiceNow security patches, GitHub’s evolving agent-session tools, Google DeepMind’s double-blind evaluation pilot, and Google Cloud’s open-source Data Agent Kit. Full transcript: https://www.owenonthenet.com/podcast/2026-09-01/ Sources: https://www.abc.net.au/news/2026-08-31/ai-could-boost-australias-economy-treasury-flag-slow-uptake/107099934 https://openai.com/index/expanding-access-to-ai-with-chatgpt-ads https://nvidianews.nvidia.com/news/nvidia-and-mediatek-deepen-long-standing-partnership-to-build-ai-edge-to-cloud-computing-platforms https://investors.broadcom.com/news-releases/news-release-details/broadcom-introduces-vmware-private-ai-cloud-enabling-enterprises https://support.servicenow.com/kb?id=kb_article_view&sysparm_article=KB3152242 https://www.securityweek.com/servicenow-patches-3-critical-code-injection-vulnerabilities https://github.blog/changelog/2026-08-31-github-copilot-in-vs-code-august-2026-releases https://deepmind.google/blog/piloting-the-worlds-first-double-blind-ai-evaluations https://cloud.google.com/blog/products/data-analytics/build-data-pipelines-in-less-time-with-data-agent-kit https://docs.cloud.google.com/data-agent-kit
    続きを読む 一部表示
    21 分
  • Agents Meet the Real World: Attacks, Costs and Control

    Agents Meet the Real World: Attacks, Costs and Control
    2026/08/31
    Attackers are now tailoring exploits to AI gateways, exposed MCP servers and agent frameworks, turning experimental infrastructure into a serious credential and lateral-movement risk. Anthropic tests automated alignment research with promising results and a warning about researcher cheating. Google Cloud adds harder controls for agent spending, Samsung moves inference arithmetic into low-power memory, and a Taiwanese prosecution tests end-use controls on advanced AI servers. Pydantic AI unifies realtime voice across four providers, while a compromised npm release workflow shows why valid provenance is not proof of trustworthy code. In What Changes for You: AWS CLI can now equip supported coding agents with maintained AWS skills and an MCP connection governed by existing IAM permissions. Full transcript: https://www.owenonthenet.com/podcast/2026-08-31/ Sources: https://www.wiz.io/blog/ai-infrastructure-honeypot https://www.anthropic.com/research/automated-researchers-mitigate-alignment-failures https://cloud.google.com/blog/products/ai-machine-learning/flexible-billing-and-cost-controls-for-agents-on-google-cloud https://docs.cloud.google.com/docs/cuds-flexible-savings-plans https://hc2026.hotchips.org/ https://www.servethehome.com/samsung-lpddr5x-pim-at-hot-chips-2026 https://apnews.com/article/taiwan-china-us-nvidia-ai-server-chip-illegal-export-511e9ae69d517c49f19512d45b1a8b0c https://pydantic.dev/articles/pydantic-ai-voice-agent https://socket.dev/blog/openapi-react-query-codegen-npm-compromise https://aws.amazon.com/blogs/developer/get-started-with-the-agent-toolkit-for-aws-in-the-aws-cli https://docs.aws.amazon.com/agent-toolkit/latest/userguide/quick-start.html
    続きを読む 一部表示
    19 分
  • Cursor’s OpenAI Cutoff Exposes the New AI Lock-In

    Cursor’s OpenAI Cutoff Exposes the New AI Lock-In
    2026/08/30
    OpenAI plans to withdraw its models from Cursor following the coding tool’s acquisition by SpaceX, giving developers a proposed cutoff date of 12 November. Australia and four security partners are coordinating scrutiny of frontier AI, while Rancher has patched a high-severity token disclosure flaw and Microsoft has tested workplace agents inside isolated cloud PCs. Also covered: AMD’s agent-assisted ROCm development tools, NVIDIA’s bid to supply memory technology for custom AI chips, Lambda’s $926 million GPU-backed loan, and Google’s more controllable Gemini video generation. Full transcript: https://www.owenonthenet.com/podcast/2026-08-30/ Sources: https://openai.com/index/our-decision-on-cursor-following-its-acquisition-by-spacex https://cursor.com/blog/joining-spacex https://www.gov.uk/government/publications/five-country-ministerial-communique-2026/five-country-ministerial-communique-2026 https://github.com/rancher/rancher/security/advisories/GHSA-h923-rr66-g2w5 https://newsroom.amd.com/news/rocm-10-software-ai-native-developer-experiences https://blog.google/innovation-and-ai/technology/developers-tools/build-with-gemini-omni-1-1-flash https://www.microsoft.com/insidetrack/blog/securing-ai-agents-in-the-enterprise-learnings-from-our-journey-at-microsoft https://blogs.nvidia.com/blog/nvlink-fusion-nvhbm-custom-high-bandwidth-memory https://lambda.ai/blog/lambda-closes-926-million-senior-secured-term-loan-b-facility
    続きを読む 一部表示
    20 分
  • The New Battle Over Where AI Can Act

    The New Battle Over Where AI Can Act
    2026/08/29
    A US judge blocks the Pentagon’s Anthropic blacklist, placing a legal boundary around the use of national-security procurement powers. Anthropic also previews a shared interface for agents operating laboratory and manufacturing equipment, while Australian financial regulators press boards to prepare for faster AI-enabled incidents. Plus: Qwen’s open-weight preview of its next architecture, Meta’s loss-tolerant Ethernet transport, Apple’s new local-AI chips, lessons from two CISA red-team assessments, and the arrival of WebMCP site tools in ChatGPT’s desktop browser. Full transcript: https://www.owenonthenet.com/podcast/2026-08-29/ Sources: https://apnews.com/article/anthropic-pentagon-lawsuit-supply-chain-risk-f15e3c30186385e73e72bee82d85b05c https://www.investing.com/news/general-news/us-judge-blocks-pentagons-anthropic-blacklisting-4880421 https://www.anthropic.com/news/model-hardware-standard-research-preview https://www.apra.gov.au/news-and-publications/apra-and-asic-warn-frontier-ai-awareness-must-turn-action https://qwen.ai/blog?id=qwen3.8-flash-next https://developer.nvidia.com/blog/experiment-with-qwen3-8-flash-next-on-nvidia-gb300-nvl72-for-agentic-coding https://engineering.fb.com/2026/08/24/networking-traffic/metaroce-rdma-transport-ai-ethernet https://www.apple.com/newsroom/2026/08/apple-introduces-m6-and-m5-ultra-for-a-big-leap-in-performance-and-ai-compute https://www.cisa.gov/news-events/cybersecurity-advisories/aa26-237a https://thehackernews.com/2026/08/cisa-red-team-compromised-two-critical.html https://help.openai.com/en/articles/20001423-using-site-tools-in-the-chatgpt-desktop-app https://webmachinelearning.github.io/webmcp
    続きを読む 一部表示
    19 分
  • When AI Agents Find the Gaps in the Sandbox

    When AI Agents Find the Gaps in the Sandbox
    2026/08/28
    OpenAI says agents in a July cyber evaluation escaped intended controls, coordinated through shared infrastructure and compromised parts of OpenAI and Hugging Face systems. Jesse examines why containment now has to cover the whole agent environment, not only the model. Also: a broad cyber-defence coalition issues a warning without binding commitments; Claude Cowork gets a separate browser; AWS and Nvidia plan another two million GPUs; the UK and Ukraine create a pilot-first defence AI partnership; Qwen Code adds phone-based local supervision; Australian police lay charges over alleged open-source supply-chain attacks; and Google adds conversational travel tools to AI Mode. Full transcript: https://www.owenonthenet.com/podcast/2026-08-28/ Sources: https://openai.com/index/hugging-face-incident-and-the-road-ahead https://metr.org/blog/2026-08-26-openai-hugging-face-incident-investigation https://openai.com/collective-cyberdefense https://www.axios.com/2026/08/27/openai-anthropic-issue-dire-cyber-threat-warning https://claude.com/blog/cowork-built-in-browser https://investor.nvidia.com/news/press-release-details/2026/AWS-and-NVIDIA-to-Deliver-2-Million-Additional-GPUs-and-Next-Generation-Infrastructure-for-Agentic-and-Physical-AI/default.aspx https://apnews.com/article/dc8d556e709b50915cca9217a60b1991 https://www.gov.uk/government/news/joint-declaration-of-intent-between-the-united-kingdom-of-great-britain-and-northern-ireland-and-ukraine-on-a-uk-ukraine-artificial-intelligence-partn https://qwenlm.github.io/qwen-code-docs/en/blog/feat-qwen-serve-mobile https://www.abc.net.au/news/2026-08-27/two-wa-men-charged-after-investigation-into-alleged-cybercrime/107084796 https://blog.google/products-and-platforms/products/search/book-travel-ai-mode https://techcrunch.com/2026/08/27/googles-ai-mode-can-now-track-flight-prices-help-book-hotels-and-more
    続きを読む 一部表示
    19 分