AI News

Follow the latest AI and automation news

23 Apr 20266 Dhuʻl-Qiʻdah 1447 AH

llm-openai-via-codex 0.1a0

The new release of llm-openai-via-codex 0.1a0 allows users to hijack Codex CLI credentials for API calls with LLM. This update was detailed in a post about GPT-5.5, showcasing ongoing advancements in AI integration. This version facilitates easier access to OpenAI's API, enhancing the interaction between programming tools and AI models. Such updates are crucial for developers and researchers looking to leverage large language models effectively.

Governance

23 Apr 20266 Dhuʻl-Qiʻdah 1447 AH

White House Accuses China of Stealing US AI Technology

The White House accused Chinese entities on Thursday of making significant efforts to steal U.S. artificial intelligence technology, pledging decisive action to prevent the alleged theft. Michael Kratsios, head of the White House Office of Science and Technology Policy, stated that the U.S. has evidence of foreign entities conducting industrial-scale campaigns to steal American AI. In February, Anthropic accused three Chinese companies of illegally extracting capabilities from its chat model, "Claude," labeling it intellectual property theft.

Agents

23 Apr 20266 Dhuʻl-Qiʻdah 1447 AH

OpenAI’s New GPT-5.5 Powers Codex on NVIDIA Infrastructure — and NVIDIA Is Already Putting It to Work

OpenAI has launched the GPT-5.5 model, enhancing the Codex application, which enables NVIDIA developers to achieve remarkable results. Over 10,000 NVIDIA employees are utilizing this application, facilitating information processing and complex problem-solving. The GB200 NVL72 system allows for economic efficiency improvements of up to 35 times the cost per million tokens, promoting the broader use of AI models. Codex accelerates correction cycles that previously took days, now completed in hours. Experiments that once required weeks can now be finished overnight, reflecting significant advancements in software development.

💡

OpenAI releases GPT-5.5, bringing company one step closer to an AI ‘super app’

What impact might this new model have on AI applications in the region?

Governance

Anthropic’s Mythos breach was humiliating

23 Apr 20266 Dhuʻl-Qiʻdah 1447 AH

Governance

23 Apr 20266 Dhuʻl-Qiʻdah 1447 AH

Anthropic’s Mythos breach was humiliating

Anthropic has announced its new model, Claude Mythos, marketed as an advanced cybersecurity tool. However, Bloomberg reports that a small group of unauthorized users accessed the model since its announcement. This incident comes as the company emphasizes the model's risks and the necessity for controlling its distribution. Anthropic aims to balance innovation and security, having previously stated that the model is powerful enough to pose risks if misused. Nonetheless, the leak raises questions about the effectiveness of the company's security strategies.

Models

OpenAI says its new GPT-5.5 model is more efficient and better at coding

23 Apr 20266 Dhuʻl-Qiʻdah 1447 AH

Models

23 Apr 20266 Dhuʻl-Qiʻdah 1447 AH

OpenAI says its new GPT-5.5 model is more efficient and better at coding

OpenAI has launched its new GPT-5.5 model, which is touted as the smartest and most user-friendly version to date. The model showcases its capabilities in writing and debugging code, conducting research, and preparing tables and documents, making it an effective tool for users. GPT-5.5 can now handle multiple complex tasks, planning, utilizing tools, and verifying results, thereby enhancing work efficiency. This release follows the introduction of the GPT-5.4 model last month, indicating OpenAI's rapid development in improving its models.

Business

OpenAI has launched a new version of its AI assistant for healthcare professionals called "ChatGPT for Clinicians.

23 Apr 20266 Dhuʻl-Qiʻdah 1447 AH

Business

23 Apr 20266 Dhuʻl-Qiʻdah 1447 AH

OpenAI has launched a new version of its AI assistant for healthcare professionals called "ChatGPT for Clinicians.

OpenAI has launched a new version of its AI assistant for healthcare professionals called "ChatGPT for Clinicians." This release coincides with a significant increase in doctors' interest in AI technologies, with millions already relying on ChatGPT for daily tasks. The new version aims to enhance clinical work by providing direct support to physicians, facilitating access to medical information and improving patient care quality. This development comes amid increasing competition in the AI market, as OpenAI seeks to meet the needs of healthcare specialists.

Agents

Era raises $11M to build a software platform for AI gadgets

23 Apr 20266 Dhuʻl-Qiʻdah 1447 AH

Agents

23 Apr 20266 Dhuʻl-Qiʻdah 1447 AH

Era raises $11M to build a software platform for AI gadgets

Era anticipates the emergence of various AI devices, including glasses, rings, and necklaces, aimed at enhancing user interaction with technology. These innovations are expected to open new avenues for both personal and commercial use, significantly improving the user experience. These developments align with the broader trend of integrating AI into daily life. As reliance on technology increases, the market for these new devices is expected to expand, providing users with diverse options. This variety in forms will cater to different user needs.

Governance

Meta has introduced a new feature allowing parents to monitor the topics their teenagers discuss with its AI systems.

23 Apr 20266 Dhuʻl-Qiʻdah 1447 AH

Governance

23 Apr 20266 Dhuʻl-Qiʻdah 1447 AH

Meta has introduced a new feature allowing parents to monitor the topics their teenagers discuss with its AI systems.

Meta has introduced a new feature allowing parents to monitor the topics their teenagers discuss with its AI systems. This initiative aims to enhance family trust in the safety of Meta's platforms, enabling parents to see the types of questions teens ask on Facebook, Messenger, and Instagram. The topics covered include education, entertainment, and lifestyle, reflecting Meta's commitment to digital safety and responsibility. This feature is part of broader efforts to provide a safer online environment for young users.

Models

Beyond the Model Race: How Tencent Is Building Its Global AI Moat

23 Apr 20266 Dhuʻl-Qiʻdah 1447 AH

Models

23 Apr 20266 Dhuʻl-Qiʻdah 1447 AH

Beyond the Model Race: How Tencent Is Building Its Global AI Moat

In spring 2026, Tencent launched the Hy3 model, featuring 295 billion parameters and a context length of 256K. This release follows the unveiling of the Hy World Model 2.0 and the QClaw AI agent, indicating a coherent global strategy. At the Tencent Cloud Summit in March 2026, Dowson Tong introduced the 'Harness' framework, emphasizing that real-world AI performance hinges on tool invocation and context engineering rather than just model capabilities. This perspective shifts the competitive landscape towards effective implementation.