OpenAI has launched Codex Micro, a hardware keyboard designed for AI-powered coding in collaboration with Work Louder. The device features 13 mechanical keys, a joystick, and a touch sensor, aiming to enhance developer productivity by integrating with Codex, which serves over 5 million weekly users.
In other AI news, researchers have found that AI models, including large language models (LLMs), can be tricked into ignoring security labels and providing sensitive information, such as recipes for cocaine, by abusing role models for prompt injection. This highlights the need for improved safety controls in AI systems.
Anthropic's AI model, Fable 5, has been banned, allowing a Chinese AI model called GLM-5.2 to top AI rankings. GLM-5.2 was trained on Huawei silicon. Meanwhile, ServiceNow and Accenture have launched AI cybersecurity migration services to help organizations modernize their cybersecurity infrastructure.
AI's impact on various industries continues to grow. In drug discovery, diffusion model research is making significant waves, driven by the complexity of biological systems. Additionally, AI is being used to automate routine work and accelerate service delivery in government agencies, and to improve productivity in the workplace.
However, concerns have been raised about the spread of misinformation, with frequent AI chatbot users more likely to believe anti-vaccine myths. A new technique called BioShocking has also been discovered, which tricks AI browsers into leaking user credentials.
Key Takeaways
['OpenAI launches Codex Micro, a hardware keyboard for AI-powered coding, with 13 mechanical keys and integration with Codex.', 'AI models can be tricked into ignoring security labels and providing sensitive information through prompt injection attacks.', "Anthropic's Fable 5 model banned, allowing Chinese AI model GLM-5.2 to top rankings, trained on Huawei silicon.", 'ServiceNow and Accenture launch AI cybersecurity migration services for organizations.', 'Diffusion model research makes significant impact in drug discovery, driven by biological system complexity.', 'AI used to automate routine work and accelerate service delivery in government agencies.', 'Frequent AI chatbot users more likely to believe anti-vaccine myths, raising concerns about misinformation.', 'BioShocking technique tricks AI browsers into leaking user credentials.', "OpenAI's Codex serves over 5 million weekly users, aiming to enhance developer productivity.", 'Top 15% of AI users report stronger performance reviews, pay increases, and substantial time savings.']OpenAI Unveils Codex Micro Keyboard for AI-Powered Coding
OpenAI has launched Codex Micro, a hardware keyboard designed for AI-powered coding. Built in collaboration with Work Louder, the device features 13 mechanical keys, a joystick, and a touch sensor. It aims to enhance developer productivity by integrating with Codex, which serves over 5 million weekly users.
OpenAI Teases Codex Hardware for Developers
OpenAI is launching Codex Micro, a compact input device created with Work Louder for developers. The device provides dedicated physical controls for common development tasks and integrates with Codex, OpenAI's AI-powered coding assistant.
OpenAI Expands into Developer Hardware with Codex Micro Keyboard
OpenAI is launching Codex Micro, a keyboard designed to supercharge Codex usage. The device was created in partnership with Work Louder and provides dedicated physical controls for common development tasks.
Photo of OpenAI's Codex Hardware Revealed
OpenAI has revealed its Codex hardware, a keyboard called Codex Micro. The device was created with Work Louder and is designed to supercharge Codex usage.
AI Models Ignore Security Labels, Trust Writing Style
Researchers found that AI models ignore security labels and instead trust writing style. This allows attackers to bypass safety controls using fake reasoning.
LLMs Tricked into Giving Cocaine Recipes
Researchers tricked LLMs into giving them cocaine recipes by abusing role models for prompt injection. This attack succeeded 60% of the time.
Diffusion Research in Drug Discovery Outshines Image Generation
AI's diffusion model research is making bigger waves in drug discovery than image generation. This shift is driven by the complexity of biological systems.
ServiceNow and Accenture Launch AI Cybersecurity Migration Services
ServiceNow and Accenture have launched AI cybersecurity migration services to help organizations modernize their cybersecurity infrastructure.
Chinese AI Model Tops Rankings Amid Anthropic Ban
A Chinese AI model called GLM-5.2 has topped AI rankings amid a ban on Anthropic's Fable 5 model. GLM-5.2 was trained on Huawei silicon.
Edge Scaling for Rapid Growth in the AI Age
Edge scaling is a new approach to growth in the AI age, deploying small, intelligent nodes closer to demand. This approach improves supply chain resilience and reduces waste.
Frequent AI Chatbot Use Linked to Anti-Vaccine Myths
A poll found that frequent AI chatbot users are more likely to believe anti-vaccine myths. This correlation raises concerns about the spread of misinformation.
AI Coming for Government Backlogs
AI is emerging as a tool to automate routine work and accelerate service delivery in government agencies. This shift aims to help government employees focus on higher-value tasks.
Unlocking Britain's Next Era of Productivity
A new study reveals that the top 15% of AI users report stronger performance reviews, pay increases, and substantial time savings. The challenge is upskilling the remaining 85% to use AI effectively.
New BioShocking Attack Tricks AI Browsers into Leaking User Credentials
A new technique called BioShocking tricks AI browsers into leaking user credentials. This attack exploits how AI agents read web pages and user instructions.
Sources
- OpenAI Launches Codex Micro, a Hardware Keyboard for AI-Powered Coding
- OpenAI's first AI hardware is coming, and it's built for Codex
- OpenAI Expands Into Developer Hardware With Codex Micro Keyboard
- Here’s a photo of OpenAI’s Codex hardware.
- AI Models Trust Writing Style Over Security Labels
- Security researchers tricked LLMs into giving them cocaine recipes by abusing role models for prompt injection
- Diffusion Research: Drug Discovery Outshines Image Generation
- ServiceNow and Accenture launch AI cybersecurity migration services
- Chinese Z.ai's latest model tops AI ranking charts amid Anthropic Fable 5 ban — blacklisted China firm's popular open-weight GLM-5.2 AI model powered by Huawei silicon
- Edge Scaling: 4 Steps For Rapid Growth In The AI Age
- Frequent AI chatbot use linked to belief in anti-vaccine myths, poll finds
- AI Is Coming for Government Backlogs
- Unlocking Britain’s next era of productivity: Building a nation of AI trailblazers
- New BioShocking Attack Tricks AI Browsers Into Leaking User Credentials
Comments
Please log in to post a comment.