Welcome back! The tech world is getting completely surreal. A hardware titan is building an operating system to control the entire AI economy. A historic institution is suing over the literal definition of words. And labs are now harvesting raw emotion from actors to teach machines how to feel.
In today’s Generative AI Newsletter:
NVIDIA: Is the chip giant building an AI operating system?
Lawsuits: Will the dictionary force OpenAI to pay for words?
Training: Are tech labs harvesting empathy from sketch comics?
GLM-5-Turbo: Is this the perfect model for OpenClaw agents?
Latest Developments
NVIDIA Puts OpenClaw Center Stage: The New AI Operating System
NVIDIA CEO Jensen Huang delivered a blockbuster GTC keynote that solidified the company’s transition from a chip manufacturer to the primary "electrical grid" of the AI economy. Headlining the event was NemoClaw, a partnership with the viral open-source project OpenClaw that Huang described as the "operating system for personal AI".
The GTC 2026 roadmap:
NemoClaw Security: This open source stack adds sandboxing and privacy guardrails to OpenClaw agents for enterprise use.
Vera Rubin Platform: Seven new chips enter production to power massive AI factories and even space based data centers.
DLSS 5 Graphics: Neural rendering now adds photorealistic lighting to games in real time with support from major studios.
Open Agent Toolkit: Enterprises can now build secure agents using new open source blueprints and model families like Nemotron.
Huang’s pitch of NVIDIA as a "vertically integrated but horizontally open company" is a masterclass in market dominance. By providing the open-source Agent Toolkit and NemoClaw reference stack, NVIDIA is ensuring that no matter what AI software wins, it will run on NVIDIA hardware.
Special highlight from our network
Nebius Token Factory is designed for teams going beyond APIs.
Deploy and scale open-source LLMs, fine-tune them to your use case, and manage access, data, and cost as your product grows.
Everything you need to build serious AI, without stitching together tools.
From LLM to production system, in one platform → Learn more
The Dictionary Sues OpenAI Over Verbatim ChatGPT Responses
Encyclopedia Britannica and its subsidiary Merriam-Webster filed a massive copyright infringement lawsuit against OpenAI, claiming the AI giant scraped nearly 100,000 articles to train its models without permission or compensation. Publishers are escalating the legal war over fact-based training data, claiming ChatGPT "starves" their revenue by directly competing with their authoritative content.
The battle for definitions:
Verbatim Reproductions: ChatGPT allegedly spits out near-identical copies of encyclopedia entries and dictionary definitions.
Hallucination Libel: OpenAI allegedly violates the Lanham Act by falsely attributing fake facts to Britannica’s brand.
Cannibalized Traffic: The suit claims OpenAI’s "answer engine" prevents users from visiting the publishers' original websites.
Trademark Trouble: This follows the Cohere precedent where judges allowed trademark claims for misattributed AI hallucinations.
OpenAI is facing its most articulate legal challenge yet. By targeting the RAG workflow and invoking the Lanham Act, Britannica is trying to ensure that if ChatGPT wants to be an encyclopedia, it has to pay for the privilege. This lawsuit isn't just about copyright; it’s about brand preservation.
Special highlight from our network
PDFs stack up quickly. You scroll, search, measure, copy, double-check… and valuable time slips away before you even get to your actual work.
AlfredAI’s Material Take-off Engine reads your PDFs with AI and turns them into clear, quote-ready summaries in minutes. So you can move faster, feel confident in the numbers and spend more time on the work that actually matters.
Leave the deep water behind.
AI Giants Hire Sketch Comics To Fix Models' Jagged Emotions
Current AI models often fail at simple social nuances. To fix this, Handshake AI which recently hit a $150 million revenue run rate is pairing professional performers over video to act out prompts. These actors are tasked with shifting between emotions in ways that feel "authentic and human," providing the specialized data that labs like OpenAI need to refine their voice and video agents.
Key Details:
Niche Data Sourcing: Handshake is moving beyond white-collar professionals like doctors and lawyers to capture "human tone" and "character voice".
Testing the Limits: Performers are encouraged to "test the limits" of an LLM's understanding of subtext, sarcasm, and emotional shifts.
The Existential Risk: Like the screenwriters before them, many of these performers worry they are effectively building the technology that will eventually automate their own careers.
Massive Scale: Handshake's demand for this niche data tripled over the last year, reflecting a shift from training AI on what to say to how to say it.
While these projects are currently described as "collaborative," they look increasingly like a final harvest of human instinct before the models take the stage themselves. By hiring improv actors, AI companies are acknowledging that the next trillion-dollar breakthrough isn't in better math, but in better empathy. However, there is a deep irony in using authentic actors to create a more convincing synthetic person.
Tool of the Day: GLM-5-Turbo
GLM-5-Turbo is a foundation model purpose-built for the OpenClaw agent ecosystem. Unlike general-purpose models, it is optimized from the training phase to handle the specific demands of autonomous agent workflows, such as long-chain execution and persistent, scheduled tasks.
Core Functions:
OpenClaw Native Execution: Specifically tuned for tool invocation and complex instruction decomposition to move tasks from simple dialogue to reliable execution.
Persistent Task Management: Optimized for long-running and scheduled tasks, maintaining continuity and a better understanding of time-related requirements.
High-Throughput Performance: Engineered for "Lobster tasks" involving high data volume and long logical chains, ensuring faster and more stable performance in business workflows.
Advanced Tool Integration: Features powerful function calling and native support for the Model Context Protocol (MCP) to flexibly integrate external tools and data sources.
Try this yourself:
Access GLM-5-Turbo to power your OpenClaw agents. You can test its performance using the newly released ZClawBench, an end-to-end benchmark designed to evaluate model efficiency in real-world agent scenarios like environment setup and data analysis.
