<?xml version="1.0" encoding="UTF-8"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>AI Titus News</title><link>https://aititus.com/news/</link><atom:link href="https://aititus.com/news/feed.xml" rel="self" type="application/rss+xml"/><description>A daily one-page pulse on AI, agentic engineering, the state of AI coding, and the wider world.</description><language>en-us</language><lastBuildDate>Tue, 25 Aug 2026 08:25:25 -0600</lastBuildDate><item><title>IT ADMINS CAN NOW SWITCH ON CLAUDE&#x27;S APP CONNECTORS FOR A WHOLE COMPANY AT ONCE. ANTHROPIC&#x27;S ENTERPRISE-MANAGED AUTH WENT LIVE TODAY THROUGH OKTA, SLACK, NOTION AND DATADOG INCLUDED</title><link>https://claude.com/blog/enterprise-managed-auth</link><guid isPermaLink="true">https://claude.com/blog/enterprise-managed-auth</guid><description>Lead story.</description><category>Headline</category><pubDate>Tue, 25 Aug 2026 08:25:25 -0600</pubDate></item><item><title>Alibaba&#x27;s Wan3.0 turns documents, slides and web pages into 30-second videos</title><link>https://technode.com/2026/08/24/alibaba-launches-wan3-0-video-model-with-30-second-generation-and-document-input/</link><guid isPermaLink="true">https://technode.com/2026/08/24/alibaba-launches-wan3-0-video-model-with-30-second-generation-and-document-input/</guid><description>Launched Monday, one day after Alibaba closed its $10.2 billion share sale. Clips run twice as long as the last version, and the model takes documents, spreadsheets, slides and web pages as input, not just text prompts. It ran in public beta since August 6 and is already used for ads, short drama and tourism videos. Feeding it a sales deck and getting a video back is the part worth testing.</description><category>AI Frontier</category><pubDate>Tue, 25 Aug 2026 08:25:25 -0600</pubDate></item><item><title>Qwen3.8-27B is the only small model in Code Arena&#x27;s WebDev top 10</title><link>https://x.com/arena/status/2091920512796725272</link><guid isPermaLink="true">https://x.com/arena/status/2091920512796725272</guid><description>It landed at #9 with 1595 points, six ranks behind its giant sibling Qwen3.8-Max. A 27B open-weights model you can run on one good GPU is now trading blows with frontier systems on real web development tasks. If you have been waiting for local models to get good enough for daily coding work, this is the signal.</description><category>AI Frontier</category><pubDate>Tue, 25 Aug 2026 08:25:25 -0600</pubDate></item><item><title>OpenAI launched AI Futures, a research blog on who holds power in an AI economy</title><link>https://openai.com/index/introducing-ai-futures/</link><guid isPermaLink="true">https://openai.com/index/introducing-ai-futures/</guid><description>Posted August 24 by OpenAI&#x27;s new Strategic Futures team. The team calls concentration of power the biggest and hardest long-run risk: how does a free society keep individual rights when a few actors control transformative AI? Planned output is posts, papers, videos and podcasts. Worth bookmarking whichever lab you favor.</description><category>AI Frontier</category><pubDate>Tue, 25 Aug 2026 08:25:25 -0600</pubDate></item><item><title>Headlong, a persistent agent that runs continuously and debugs itself for $1 to $2 an hour</title><link>https://x.com/andykonwinski/status/2091990178638496195</link><guid isPermaLink="true">https://x.com/andykonwinski/status/2091990178638496195</guid><description>Andy Konwinski showed an agent that keeps running instead of stopping at the end of a task, fixing its own failures along the way, with a 48-minute self-debugging run as the demo. The always-on agent that watches its own work is becoming a product category, and the hourly cost is already close to nothing.</description><category>AI Frontier</category><pubDate>Tue, 25 Aug 2026 08:25:25 -0600</pubDate></item><item><title>An $800 mining GPU was unlocked to 64GB and runs Qwen3.8-27B at 84 tokens a second</title><link>https://github.com/amoghmunikote/cmpunlocker</link><guid isPermaLink="true">https://github.com/amoghmunikote/cmpunlocker</guid><description>A kernel patch unlocks the full 64GB of HBM on NVIDIA&#x27;s CMP 170HX, a card sold cheap for crypto mining. The author reports 84 tokens a second at short context and 57 at 200K, with prefix caching cutting a 200K prefill from 157 seconds to 2.5. Numbers are the author&#x27;s own, but the repo is public and the recipe is documented.</description><category>AI Frontier</category><pubDate>Tue, 25 Aug 2026 08:25:25 -0600</pubDate></item><item><title>affaan-m/ECC</title><link>https://github.com/affaan-m/ECC</link><guid isPermaLink="true">https://github.com/affaan-m/ECC</guid><description>The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor </description><category>What&#x27;s Being Built</category><pubDate>Tue, 25 Aug 2026 08:25:25 -0600</pubDate></item><item><title>NousResearch/hermes-agent</title><link>https://github.com/NousResearch/hermes-agent</link><guid isPermaLink="true">https://github.com/NousResearch/hermes-agent</guid><description>The agent that grows with you</description><category>What&#x27;s Being Built</category><pubDate>Tue, 25 Aug 2026 08:25:25 -0600</pubDate></item><item><title>deepseek-ai/deepseek-harness</title><link>https://github.com/deepseek-ai/deepseek-harness</link><guid isPermaLink="true">https://github.com/deepseek-ai/deepseek-harness</guid><description>DeepSeek Harness: Everything is a Plugin.</description><category>What&#x27;s Being Built</category><pubDate>Tue, 25 Aug 2026 08:25:25 -0600</pubDate></item><item><title>SWE Refactor Bench: Can Coding Agents Complete a Long-Horizon, Whole-Repository Stack Migration?</title><link>http://arxiv.org/abs/2608.23564v1</link><guid isPermaLink="true">http://arxiv.org/abs/2608.23564v1</guid><description>Research paper.</description><category>Research</category><pubDate>Tue, 25 Aug 2026 08:25:25 -0600</pubDate></item><item><title>Prime Agent: A Self-Improving RLM Harness</title><link>http://arxiv.org/abs/2608.23552v1</link><guid isPermaLink="true">http://arxiv.org/abs/2608.23552v1</guid><description>Research paper.</description><category>Research</category><pubDate>Tue, 25 Aug 2026 08:25:25 -0600</pubDate></item><item><title>When &quot;Do Not&quot; Is Not Deny: Security Rules in CLAUDE.md vs Built-In Controls</title><link>http://arxiv.org/abs/2608.23550v1</link><guid isPermaLink="true">http://arxiv.org/abs/2608.23550v1</guid><description>Research paper.</description><category>Research</category><pubDate>Tue, 25 Aug 2026 08:25:25 -0600</pubDate></item></channel></rss>