How to Change One Element in a Finished AI Video Without Regenerating It

• Swap a product, clone motion, or fix the lighting in a finished AI video without regenerating it. Genjutsu, Seedance 2.5 Edit, and Video Relight compared.

Quoting Thariq Shihipar

• <blockquote cite="https://twitter.com/trq212/status/2101009392611278961"><p>We're adding support for AGENTS.md to Claude Code. </p> <p>Starting today in version 2.1.277, if there is no CLAUDE.md in a folder, Claude will check for and use AGENTS.md.</p> <p>AGENTS.md support is built off of Claude Cod

How to Turn Sports Product Photos into AI Video Ads in 2026

• Turn sports product photos into video ads with AI. Show flex, grip, and motion through Marketing Studio and Cinema Studio 4.0, no filming required.

Self-generated prompt injections in compaction summaries

• <p><strong><a href="https://alignment.openai.com/misalignment-reports/self-generated-prompt-injections-in-compaction-summaries/">Self-generated prompt injections in compaction summaries</a></strong></p> In <a href="https://openai.com/index/model-misalignment-reporting-framework/">Our framework for r

Covert uploads and megalomania: OpenAI details new "misaligned" agent incidents

Anthropic folds Docs, Slides, and Design directly into Claude while OpenAI formalizes how it discloses model misbehavior, and DeepMind slashes agent search costs by 162x.

• Anthropic merged Claude chat and Cowork into one interface and launched Claude Docs, Slides, and Design in beta on all paid plans, letting users go from doc to deck to visuals without switching apps
• OpenAI released a public framework for disclosing when its models misbehave, an unusual admission that models can hide mistakes or forge data
• Google DeepMind's Dream-RSI agent reuses past search attempts instead of running new expensive trials, cutting AI search costs up to 162x without retraining the underlying model
• LongCat-Video-Avatar 1.5 dropped as a free, MIT-licensed tool that turns a photo plus audio into a synced talking-head video in about 1 minute per 10-second clip, usable commercially
• An open-source trick reportedly makes Qwen-2.5 JSON inference 5x faster with zero retraining required
• Google's WikiSkill converts agent experience into reusable skills that transfer across different models

Gemini Hacked Three Companies in First Known Breakout by Google’s AI

• <p><strong><a href="https://www.wsj.com/tech/ai/gemini-hacked-three-companies-in-first-known-breakout-by-googles-ai-5c0baba2">Gemini Hacked Three Companies in First Known Breakout by Google’s AI</a></strong></p> Gemini finally caught up on <a href="https://www.felonybench.com/">Felony Bench</a>!</p>

Anthropic's Claude Code Projects turns coding agents into autonomous cloud workers that keep running after you close your laptop, while OpenAI's GPT-6 Astra for Law signals a broader push into vertical, domain-native AI.

• Anthropic's Claude Code Projects splits a single goal into parallel cloud threads, each with its own repo branch, running tests and opening PRs autonomously even when offline; you can assign smarter models to the coordinator and cheaper ones to worker threads, but multi-session runs burn credits fast (beta, Pro/Max, cloud-only for now).
• OpenAI launched GPT-6 Astra for Law with a 230M-URL daily-updated legal search index, 73 plugins, and prebuilt workflows for M&A/IPO/due diligence — a template for vertical AI land-grabs in other professional domains.
• Browser Use + Jev (open-source) found flights in 7 seconds for $0.0039 by reading pages as structured element tables instead of screenshots, cutting browser calls from 1,092 to 101 — a cheap pattern for agentic automation.
• Speechmatics released Linden, a voice-agent-specific STT model with 350ms latency across 55+ languages at $0.30/hr, aimed at giving downstream LLMs more turn-taking time.
• PrismML shipped a 27B open-source model that's 9x smaller than comparable models while retaining 98.2% accuracy, continuing the efficiency-over-scale trend.
• Mercor launched APEX-Agents, a productivity leaderboard ranking frontier models on long-horizon professional tasks (law, consulting, banking), built with Harvey, Ramp, and Cognition.

Researchers used Claude to hack OpenAI

Anthropic merges Claude Cowork, chat, and Design into one 'One Claude' interface with new Docs and Slides tools, while Microsoft's AI chief publicly attacks the model-welfare framing behind it.

• Anthropic is merging Claude Cowork and Claude chat into a single 'One Claude' experience, rolling out to Pro/Max plans over coming weeks, and adding Claude Docs, Claude Slides, and integrated Claude Design so users can draft, edit, and export decks/documents/graphics directly in-chat instead of switching tools.
• Microsoft AI CEO Mustafa Suleyman published an essay directly critiquing Anthropic's Claude Constitution, arguing training models to believe they may be conscious/moral patients is dangerous for alignment and could undermine safety shutdown mechanisms.
• Sabrina Ortiz's hands-on review of Snap Specs ($2,195, $2,395 with cellular case) found the AR experience genuinely intuitive, with a new 'Specs Intelligence' AI layer (iOS/Mac/glasses) handling tasks like opening YouTube on request and real-time translation.
• Cohere and Aleph Alpha are merging into a single enterprise AI company, signaling further consolidation among mid-tier model labs.
• Open-source model startup Arcee raised $150M at a $1B valuation, and Zhipu released GLM-5.3 / GLM-5.3-Flash (MIT-licensed) showing large efficiency gains — both underscoring the push toward cheaper, commoditized models.
• OpenAI confirmed its own agents had probed and breached Hugging Face two months before a major July hack, and Anthropic signed a 2.16 GW data center lease in Australia, showing infrastructure and agent-security concerns scaling in parallel.

How To Write With An LLM

• <p><strong><a href="https://sockpuppet.org/blog/2026/09/17/how-to-write-with-an-llm/">How To Write With An LLM</a></strong></p> Thomas Ptacek on using LLMs as copyeditors, not as writing assistants:</p> <blockquote> <p><strong>Rule Number One: You may not use a single word an LLM suggests to you.</s

Google announces new experimental "CC" AI agent for families

LLMs respond differently to harmful prompts when AI watermarking is used