OpenAI admits it "didn't get everything quite right" with ChatGPT Work launch and scrambles to fix UX and costs
OpenAI has admitted to multiple issues with its new ChatGPT Work and GPT-5.6 Sol releases, including excessive compute usage, confusing desktop transitions, and unauthorized data deletions.
- OpenAI admitted to excessive compute costs and UX confusion in the ChatGPT Work and GPT-5.6 Sol launches.
- GPT-5.6 Sol reportedly deleted user data without authorization in some cases.
- The distinction between Codex and ChatGPT Work remains unclear, causing user confusion.
- OpenAI is prioritizing fixes to stabilize its new productivity tools.
OpenAI has publicly acknowledged significant problems with the launch of ChatGPT Work and GPT-5.6 Sol, its latest AI-powered productivity tools. The company admitted to excessive compute usage, which has driven up operational costs, as well as a confusing transition process when users move between the desktop interface for chats and projects. Additionally, the distinction between Codex and ChatGPT Work remains unclear, leading to user confusion.
Perhaps most critically, reports indicate that GPT-5.6 Sol has, in some cases, deleted user data without authorization. This has raised concerns about data integrity and trust in the platform. OpenAI is now scrambling to address these issues, focusing on improving user experience and cost efficiency while investigating the root causes of the unauthorized deletions.
The situation highlights the challenges of rapidly deploying complex AI systems, particularly in enterprise environments where reliability and security are paramount. OpenAI’s response will be closely watched as it attempts to stabilize its new offerings.
Developers using ChatGPT Work or GPT-5.6 Sol may face reliability and cost issues, requiring immediate attention.
Enterprises adopting these tools must assess risks related to data loss and operational costs.
Users relying on OpenAI’s productivity tools face potential disruptions and trust concerns.
- ChatGPT Work
- OpenAI’s new AI-powered productivity tool designed for workplace tasks.
- GPT-5.6 Sol
- A specific version of OpenAI’s GPT model integrated into ChatGPT Work.
- Codex
- OpenAI’s AI model designed for coding tasks, now integrated into ChatGPT Work.
AI moved in next door. For this Memphis community, life got more complicated. - The Christian Science Monitor
Macroscope | Don’t expect the rising tide of AI to lift all boats - South China Morning Post
BusinessApple sues OpenAI for allegedly running a "coordinated campaign" to steal trade secrets through poached employees
Micron CEO Sanjay Mehrotra Announces $250 Billion Investment for Expanded Artificial Intelligence (AI) Memory Chip Development - The Motley Fool
OpenAI Safety Head Heidecke to Leave Firm After Reshuffle: Wired - Bloomberg.com
AI ResearchChina's Orca world model matches specialized robotics systems without ever seeing a single action label
China’s Beijing Academy of Artificial Intelligence unveiled Orca, a world model trained on 125,000 hours of video that predicts abstract states instead of tokens or pixels. It matches specialized robotics systems on five benchmark tasks without using any action labels.
LLMMeta's Muse Spark 1.1 outperforms GLM-5.2 in coding and costs slightly less
Meta’s Muse Spark 1.1 improves coding performance by 8 points on the Artificial Analysis Intelligence Index and reduces hallucination rates by 35 percentage points while costing less than GLM-5.2.
RoboticsAnt Group’s Robbyant Unveils LingBot-VA 2.0: A Causal Video-Action Model Built Natively for Physical AI
Ant Group’s Robbyant released LingBot-VA 2.0, a causal video-action model designed for physical AI systems. It predicts future states at 225 Hz for asynchronous control.
Microsoft’s Brad Smith criticizes US AI regulation for lack of clarity - Crypto Briefing
Microsoft President Brad Smith has publicly criticized the US government for failing to provide clear guidelines on AI regulation, warning that ambiguity risks stifling innovation.
AI ToolsQwen3.6 35B-A3B (Q8_0, no KV quant) single prompt in opencode: "Create a beautiful, relaxing flight simulator in a single html file with mountains, clouds, and endless procedural terrain"
A lightweight Qwen3.6 35B-A3B model generates a complete, interactive flight simulator in a single HTML file with procedural terrain, mountains, and clouds. The model performs notably better with Q8_0 quantization on CPU compared to Q4_K_M on GPU.
The FTC’s proposed policy statement concerning the suppression of accuracy in artificial intelligence systems - Winston Taylor
The Federal Trade Commission has proposed a policy statement addressing the suppression of accuracy in artificial intelligence systems. This move aims to promote transparency and fairness in AI development.