{ "schemaVersion": 2, "slug": "2026-08-28", "title": "AI Round-up: 28 August 2026", "date": "2026-08-28", "publishedAt": "2026-08-28T08:02:35Z", "sinceUtc": "2026-08-27T07:24:15Z", "rangeLabel": "Last 7 days", "summary": "Anthropic opened a research preview of its Model Hardware Standard for safer AI interaction with physical equipment, while OpenAI joined Anthropic, AWS, Google, Microsoft and Oracle in calling for a global surge in cyber defence. Google DeepMind also began piloting double-blind evaluations intended to reduce bias in frontier-model comparisons.\n\nElsewhere, Gemini extended Omni 1.1 Flash video generation, Cohere released Parse 5, Microsoft introduced Agent Hooks and Replit added intelligent model routing. A judge blocked the Pentagon’s Anthropic blacklist, Nvidia paused new cloud-computing deals, and researchers linked Cursor’s AI coding tools to a cybercrime campaign.", "leadSignal": "Anthropic’s Model Hardware Standard", "topSignals": [ { "label": "Anthropic’s Model Hardware Standard", "href": "https://www.anthropic.com/news/model-hardware-standard-research-preview", "tag": "News" }, { "label": "OpenAI’s global cyber-defence call", "href": "https://x.com/OpenAI/status/2093074192636018977", "tag": "X" }, { "label": "DeepMind’s double-blind frontier evaluations", "href": "https://deepmind.google/blog/piloting-the-worlds-first-double-blind-ai-evaluations/", "tag": "News" } ], "topics": [ "Anthropic", "Google / Gemini", "Microsoft / Copilot", "AI agents", "AI coding tools" ], "totalSignals": 40, "stats": { "news": 10, "x": 10, "youtube": 10, "reddit": 10 }, "paths": { "html": "/roundup/2026-08-28/", "json": "/roundup/2026-08-28.json", "markdown": "/roundup/2026-08-28.md" }, "urls": { "html": "https://www.ai-roundup.dev/roundup/2026-08-28/", "json": "https://www.ai-roundup.dev/roundup/2026-08-28.json", "markdown": "https://www.ai-roundup.dev/roundup/2026-08-28.md" }, "sections": [ { "id": "news", "title": "News / Articles", "shortLabel": "News", "mobileTabLabel": "News", "count": 10, "items": [ { "title": "Piloting the world's first double-blind AI evaluations", "source": "Google DeepMind", "publishedAt": "2026-08-27T12:59:37Z", "url": "https://deepmind.google/blog/piloting-the-worlds-first-double-blind-ai-evaluations/" }, { "title": "Fast, fault-tolerant PyTorch training on AI Runtime", "source": "Databricks", "publishedAt": "2026-08-28T01:44:05Z", "url": "https://www.databricks.com/blog/fast-fault-tolerant-pytorch-training-ai-runtime" }, { "title": "Expanding our support for scientists", "source": "Anthropic", "publishedAt": "2026-08-27T19:28:34Z", "url": "https://www.anthropic.com/news/expanding-support-for-scientists" }, { "title": "Agent Hooks: An open, framework-neutral AI governance contract", "source": "Microsoft", "publishedAt": "2026-08-27T17:22:54Z", "url": "https://commandline.microsoft.com/agent-hooks-framework-neutral-ai-governance-contract/" }, { "title": "Previewing the Model Hardware Standard", "source": "Anthropic", "publishedAt": "2026-08-27T17:58:11Z", "url": "https://www.anthropic.com/news/model-hardware-standard-research-preview" }, { "title": "Gemini Omni 1.1 Flash lets you build with more control", "source": "Google DeepMind", "publishedAt": "2026-08-27T16:34:19Z", "url": "https://deepmind.google/blog/gemini-omni-1-1-flash-lets-you-build-with-more-control/" }, { "title": "Anthropic unveils new framework allowing AI agents to operate physical devices", "source": "Reuters", "publishedAt": "2026-08-27T19:20:09Z", "url": "https://www.reuters.com/technology/anthropic-unveils-new-framework-allowing-ai-agents-operate-physical-devices-2026-08-27/" }, { "title": "US judge blocks Pentagon's Anthropic blacklisting", "source": "Reuters", "publishedAt": "2026-08-28T01:27:00Z", "url": "https://www.reuters.com/legal/government/us-judge-blocks-pentagons-anthropic-blacklisting-2026-08-28/" }, { "title": "Major tech companies call for defensive surge to defeat AI-driven hacks", "source": "Reuters", "publishedAt": "2026-08-27T18:49:13Z", "url": "https://www.reuters.com/legal/litigation/major-tech-companies-call-defensive-surge-defeat-ai-driven-hacks-2026-08-27/" }, { "title": "EXCLUSIVE: Russian-speaking cybercriminals used SpaceX’s Cursor AI tool to hack seven companies", "source": "Reuters", "publishedAt": "2026-08-27T10:03:00Z", "url": "https://www.reuters.com/world/russian-speaking-cybercriminals-used-spacexs-cursor-ai-tool-hack-seven-companies-2026-08-27/" } ] }, { "id": "youtube", "title": "YouTube", "shortLabel": "Video", "mobileTabLabel": "YouTube", "count": 10, "items": [ { "title": "Sam Altman :‘AGI in 2026’, just as Models Start to [Mis]Train Themselves", "channel": "AI Explained", "publishedAt": "2026-08-27T19:16:24Z", "views": 79657, "url": "https://www.youtube.com/watch?v=KL9_1GbmCic" }, { "title": "Google takes am at Anthropic, Microsoft with budget-friendly AI pricing", "channel": "CNBC Television", "publishedAt": "2026-08-27T16:30:38Z", "views": 40561, "url": "https://www.youtube.com/watch?v=uEpbrIoftq4" }, { "title": "How I turn endless research tabs into insightful stories with Gemini in Chrome!", "channel": "Aevy TV", "publishedAt": "2026-08-27T12:55:52Z", "views": 29651, "url": "https://www.youtube.com/watch?v=x507CvZnN34" }, { "title": "Codex vs Claude - an Honest Comparison", "channel": "Tech With Tim", "publishedAt": "2026-08-27T14:00:23Z", "views": 25177, "url": "https://www.youtube.com/watch?v=sbuwHi1Sz6k" }, { "title": "How Cursor Built One of AI’s Fastest-Growing Companies", "channel": "a16z", "publishedAt": "2026-08-27T14:30:08Z", "views": 12501, "url": "https://www.youtube.com/watch?v=GHrnbvkVPZA" }, { "title": "Nvidia’s AI Boom, Salesforce’s Anthropic Bet | Bloomberg Tech 8/27/2026", "channel": "Bloomberg Tech", "publishedAt": "2026-08-27T18:43:21Z", "views": 11854, "url": "https://www.youtube.com/watch?v=H7Pjv8O4f4E" }, { "title": "Nvidia AI Boom, Meta’s Layoffs, OpenAI Rogue Agents: Techno Tyrants of Silicon | Vantage", "channel": "Firstpost", "publishedAt": "2026-08-27T17:24:07Z", "views": 10463, "url": "https://www.youtube.com/watch?v=Sf4pTk9_BLc" }, { "title": "Anthropic-Salesforce deal eases doomsday software fears", "channel": "CNBC Television", "publishedAt": "2026-08-27T18:16:53Z", "views": 9607, "url": "https://www.youtube.com/watch?v=rKtn-R5BVD4" }, { "title": "How To Build AI Software In 2026", "channel": "Erik Cupsa", "publishedAt": "2026-08-27T20:06:27Z", "views": 6082, "url": "https://www.youtube.com/watch?v=AONmDtSHlZE" }, { "title": "VS Code Live: Latest News, Vercel Plugin + New Feature Demos from VS Code Engineers", "channel": "Visual Studio Code", "publishedAt": "2026-08-28T05:36:05Z", "views": 5309, "url": "https://www.youtube.com/watch?v=ZUrQxvMoWrY" } ] }, { "id": "reddit", "title": "Reddit", "shortLabel": "Reddit", "mobileTabLabel": "Reddit", "count": 10, "items": [ { "title": "Title: Cursor account permanently terminated 12 days after upgrading to Pro+ — no explanation", "subreddit": "cursor", "publishedAt": "2026-08-27T17:02:35Z", "score": 13, "comments": 27, "url": "https://www.reddit.com/r/cursor/comments/1vzzdda/title_cursor_account_permanently_terminated_12/" }, { "title": "Claude, Codex, and Hermes installed unowned code inside corporate networks", "subreddit": "ciso", "publishedAt": "2026-08-27T21:02:35Z", "score": 8, "comments": 2, "url": "https://www.reddit.com/r/ciso/comments/1w07054/claude_codex_and_hermes_installed_unowned_code/" }, { "title": "OpenAI in panic after security breaches from autonomous AI models: ---- \"We have a limited window to strengthen cyber defenses, and with organizations including @AnthropicAI, @Google, @Microsoft, we're calling for a global effort\"", "subreddit": "singularity", "publishedAt": "2026-08-28T07:02:35Z", "score": 6, "comments": 14, "url": "https://www.reddit.com/r/singularity/comments/1w0j8cp/openai_in_panic_after_security_breaches_from/" }, { "title": "GitHub Copilot inline suggestions stop working when the parent folder name contains a space in VS Code", "subreddit": "GithubCopilot", "publishedAt": "2026-08-27T21:02:35Z", "score": 5, "comments": 1, "url": "https://www.reddit.com/r/GithubCopilot/comments/1w06juw/github_copilot_inline_suggestions_stop_working/" }, { "title": "Vivado AI Assistant — anyone actually got access? (2026.1 EA, VS Code + Copilot, MCP-based)", "subreddit": "FPGA", "publishedAt": "2026-08-28T07:29:35Z", "score": 3, "comments": 5, "url": "https://www.reddit.com/r/FPGA/comments/1w0juvv/vivado_ai_assistant_anyone_actually_got_access/" }, { "title": "Does Codex lose track of long-running terminal commands for anyone else? I have a workaround", "subreddit": "codex", "publishedAt": "2026-08-28T07:29:35Z", "score": 3, "comments": 2, "url": "https://www.reddit.com/r/codex/comments/1w0juif/does_codex_lose_track_of_longrunning_terminal/" }, { "title": "Anthropic's Model Hardware Standard Enables AI Control of Physical Devices", "subreddit": "Agent_AI", "publishedAt": "2026-08-28T07:02:35Z", "score": 2, "comments": 1, "url": "https://www.reddit.com/r/Agent_AI/comments/1w0jdna/anthropics_model_hardware_standard_enables_ai/" }, { "title": "Codex CLI 0.150.0 -> 0.150.1 (task mentions, smarter `/copy`, auto titles, Interrupt hooks + image-aware compaction)", "subreddit": "CodexAutomation", "publishedAt": "2026-08-27T15:02:35Z", "score": 2, "comments": 0, "url": "https://www.reddit.com/r/CodexAutomation/comments/1vzwebs/codex_cli_01500_01501_task_mentions_smarter_copy/" }, { "title": "Claude Code and Codex can share a repo. That is not the same as sharing a task.", "subreddit": "AgentsOfAI", "publishedAt": "2026-08-28T07:02:35Z", "score": 1, "comments": 2, "url": "https://www.reddit.com/r/AgentsOfAI/comments/1w0j60p/claude_code_and_codex_can_share_a_repo_that_is/" }, { "title": "AI code compiles almost every time and is secure about half the time. That gap is worse in Cursor agent mode.", "subreddit": "cursor", "publishedAt": "2026-08-28T06:02:35Z", "score": 1, "comments": 2, "url": "https://www.reddit.com/r/cursor/comments/1w0ie2w/ai_code_compiles_almost_every_time_and_is_secure/" } ] }, { "id": "x", "title": "X / Twitter", "shortLabel": "X", "mobileTabLabel": "X", "count": 10, "items": [ { "text": "OpenAI joined Anthropic, AWS, Google, Microsoft and Oracle in calling for a global effort to give cyber defenders the tools, resources and support needed to protect critical infrastructure.", "handle": "OpenAI", "publishedAt": "2026-08-27T20:32:29Z", "views": 2000000, "likes": 11000, "url": "https://x.com/OpenAI/status/2093074192636018977" }, { "text": "Anthropic opened the first research-preview phase of Model Hardware Standard, a new standard intended to let AI agents operate physical equipment safely in scientific research and advanced manufacturing.", "handle": "AnthropicAI", "publishedAt": "2026-08-27T18:10:22Z", "views": 1600000, "likes": 7600, "url": "https://x.com/AnthropicAI/status/2093038426140651791" }, { "text": "Google DeepMind is piloting double-blind evaluations for frontier AI in a secure environment where neither test prompts nor model weights are revealed to external evaluators.", "handle": "GoogleDeepMind", "publishedAt": "2026-08-27T13:05:44Z", "views": 275000, "likes": 640, "url": "https://x.com/GoogleDeepMind/status/2092961763553677387" }, { "text": "Gemini Live added agentic voice features for listening to a Daily Brief, triaging email and getting help from Spark.", "handle": "GeminiApp", "publishedAt": "2026-08-27T18:01:47Z", "views": 85000, "likes": 933, "url": "https://x.com/GeminiApp/status/2093036268288610520" }, { "text": "Cohere introduced Parse 5, a document-parsing model aimed at high-volume enterprise work with improved price-performance and low per-page costs.", "handle": "Cohere", "publishedAt": "2026-08-27T13:08:16Z", "views": 70000, "likes": 743, "url": "https://x.com/Cohere/status/2092962399754055863" }, { "text": "GitHub profiled Peter Steinberger’s experience maintaining OpenClaw, which it described as the fastest-growing project in GitHub history.", "handle": "github", "publishedAt": "2026-08-27T16:10:53Z", "views": 61000, "likes": 215, "url": "https://x.com/github/status/2093008357644779643" }, { "text": "Gemini Omni 1.1 Flash can now extend uploaded or generated video scenes to create longer stories or branch into new creative directions.", "handle": "GeminiApp", "publishedAt": "2026-08-27T22:45:44Z", "views": 48000, "likes": 650, "url": "https://x.com/GeminiApp/status/2093107727363092677" }, { "text": "Google DeepMind discussed how teaching AI systems self-doubt and probability could produce safer and more reliable decisions in domains from weather forecasting to robotics.", "handle": "GoogleDeepMind", "publishedAt": "2026-08-27T16:55:40Z", "views": 42000, "likes": 356, "url": "https://x.com/GoogleDeepMind/status/2093019627437662680" }, { "text": "Replit introduced Intelligent Model Routing to select the best model for each task automatically, claiming top-tier output at up to 65% lower cost.", "handle": "Replit", "publishedAt": "2026-08-27T16:21:48Z", "views": 10000, "likes": 61, "url": "https://x.com/Replit/status/2093011104884560121" }, { "text": "Together AI made GLM-5.3 Flash available, describing Z.ai’s multimodal 320B-parameter model as near GPT-5.6 Luna performance on DeepSWE at more than twice the work per budget.", "handle": "togethercompute", "publishedAt": "2026-08-27T16:38:18Z", "views": 10000, "likes": 151, "url": "https://x.com/togethercompute/status/2093015257560281099" } ] } ] }