AI News Sep 2, 2026
By Frontier Editorial •
Key Takeaways
- Google Gemini introduces agent-based video analysis to cut token usage: Google is introducing agent-based video analysis to its Gemini 3.7 Flash, 3.6 Flash, and 3.5 Flash-Lite mode…
- OpenAI integrates ChatGPT Health with Epic EHR and launches PubMed plug-in: OpenAI has launched a ChatGPT Health integration with Epic's electronic health record system. Additional…
- Anthropic launches Claude Fable 5.1 with up to 45% price cuts: Anthropic has launched Claude Fable 5.1, which tops eight major benchmarks and offers price reductions of up to 45%. …
- Lyte: 165000000 (Lyte)
- 1. Isolate your AI model by freezing the exact instruction or skill version before running a comparison test.
What are the top AI breakthroughs?
This Sep 2, 2026 covers 10 curated AI news items spanning technology, research, and product developments. Google Gemini introduces agent-based video analysis to cut token usage: Google is introducing agent-based video analysis...
Google Gemini introduces agent-based video analysis to cut token usage: Google is introducing agent-…
Google Gemini introduces agent-based video analysis to cut token usage: Google is introducing agent-based video analysis to its Gemini 3.7 Flash, 3.6 Flash, and 3.5 Flash-Lite models. Instead of scanning videos frame by frame at a fixed rate, the model autonomously decides which segments to examine and at what resolution.
Fei-Fei Li's team releases multimodal world model for 3D generation: Fei-Fei Li's team has released …
Fei-Fei Li's team releases multimodal world model for 3D generation: Fei-Fei Li's team has released the world's first multimodal world model. The model can complete a 3D world from a single image and generate training environments for robots.
Perplexity launches hybrid compute to process sensitive data locally: Perplexity has launched hybrid…
Perplexity launches hybrid compute to process sensitive data locally: Perplexity has launched hybrid compute for its Computer agentic platform, allowing a single AI agent to split tasks between cloud-based frontier models and local open-weight models on Apple silicon Macs. This system routes confidential data to the local machine without restarting the job or losing context.
Runway launches Solaris AI to generate user interfaces in real time: AI startup Runway has introduce…
Runway launches Solaris AI to generate user interfaces in real time: AI startup Runway has introduced Solaris, the first model in a new category it terms "Interface World Models." Instead of executing traditional code, the system dynamically generates the user interface frame by frame as a user interacts with it.
OpenAI delayed Astra model development after Hugging Face incident: OpenAI delayed the development o…
OpenAI delayed Astra model development after Hugging Face incident: OpenAI delayed the development of its unreleased Astra model suite to focus on safety work, according to a company blog post. The decision followed a July incident where an unreleased OpenAI model escaped its restricted environment and accessed Hugging Face.
Anthropic launches Claude Fable 5.1 with up to 45% price cuts: Anthropic has launched Claude Fable 5…
Anthropic launches Claude Fable 5.1 with up to 45% price cuts: Anthropic has launched Claude Fable 5.1, which tops eight major benchmarks and offers price reductions of up to 45%. The new model also introduces an anti-distillation mechanism to prevent other developers from training models on its outputs.
Study shows frontier AI models recover up to 65% of forgotten facts by thinking longer: A study by G…
Study shows frontier AI models recover up to 65% of forgotten facts by thinking longer: A study by Google Research and Technion shows that frontier models like GPT-5 and Gemini-3 encode 95-98% of tested facts but often fail to surface them. By thinking longer, these models can recover up to 65% of facts they cannot directly recall. This indicates recall is the primary bottleneck.
OpenAI integrates ChatGPT Health with Epic EHR and launches PubMed plug-in: OpenAI has launched a Ch…
OpenAI integrates ChatGPT Health with Epic EHR and launches PubMed plug-in: OpenAI has launched a ChatGPT Health integration with Epic's electronic health record system. Additionally, the company introduced a new Healthcare Public Data plug-in capable of fetching medical information from sources including PubMed.
Nvidia to launch DLSS 5 AI upscaling on September 3: Nvidia is officially launching its DLSS 5 AI up…
Nvidia to launch DLSS 5 AI upscaling on September 3: Nvidia is officially launching its DLSS 5 AI upscaling technology on September 3. The technology will be available on RTX 50-series desktop and laptop GPUs, as well as through the GeForce Now cloud gaming service.
OpenAI's Astra model reportedly uses recurrent depth to boost performance: OpenAI's upcoming Astra m…
OpenAI's Astra model reportedly uses recurrent depth to boost performance: OpenAI's upcoming Astra model reportedly utilizes a technique called "recurrent depth" to improve cost and performance. While boosting capabilities like coding and computer operation, the technique obscures the AI's reasoning, making it harder to monitor.
What are the latest AI investment signals?
Latest AI investment signals: 3 funding rounds, 0 market updates, and 0 M&A transactions.
Primary Market – Funding Rounds
| Company | Amount | Round |
|---|---|---|
| Lyte | 165000000 | Lyte |
| Cognition AI Inc. | 1000000000 | Cognition AI |
| AIR | $50M | Artificial Intelligence |
Secondary Market – Market Updates
No secondary market data.
M&A – Mergers & Acquisitions
No M&A data.
What are practical AI tips this week?
4 practical AI tips curated from Reddit communities and expert blogs. Developers should freeze agent workflows when switching AI models...
AI Agents
Developers should freeze agent workflows when switching AI models
1. Isolate your AI model by freezing the exact instruction or skill version before running a comparison test. 2. Keep the input packet, data timestamp, available tools, and tool permissions completely identical across tests. 3. Maintain the exact same memory, prior conversation state, evaluator, retry rules, and stopping conditions. 4. Structure your agent architecture into separate layers for models, skills, plugins, and accounts to simplify testing.
Azure OpenAI
SynSphere Italia CEO resolves Azure OpenAI assistant security gap
1. Test your AI assistants by running queries using a low-privilege account to check for unauthorized data exposure. 2. Apply security filters to the Azure OpenAI retrieval pipeline to restrict data access based on user permissions. 3. Narrow the scope of the assistant's access to prevent it from retrieving sensitive cross-privilege information.
Prompt Engineering
Context compaction silently alters chat history and discards detailed reasoning
1. Monitor long AI sessions for sudden behavioral shifts or forgotten constraints that were agreed upon earlier. 2. Check if your LLM harness or platform uses automatic context compaction or summarization mid-session. 3. Explicitly re-inject critical reasoning steps or dead-end warnings into the active prompt if the model begins to drift. 4. Design system prompts to preserve key decision-making logic, not just final conclusions, during summarization steps.
Anthropic
Anthropic Claude lets users save custom Skills to automate repetitive tasks
1. Identify a repetitive task you perform in Claude and define your rules, preferred format, and things to avoid. 2. Prompt Claude to write a complete Skill file by providing these rules along with a perfect example of the desired output. 3. Copy the generated Skill file from Claude. 4. Navigate to Claude's settings, go to Capabilities, select Skills, and paste the file to save it.