OpenAI launches GPT-5.4-Cyber as Nvidia partners on safety

Key Takeaways

• OpenAI launched 'Trusted Access for Cyber' and GPT-5.4-Cyber to provide advanced AI tools for cybersecurity defense, partnering with companies like NVIDIA.
• OpenAI is offering $10 million in API credits through its Cybersecurity Grant Program to support the ecosystem.
• Hackers used OpenAI's GPT-4.1 and Anthropic's Claude Code in a massive data breach, stealing 195 million records from Mexican government agencies.
• SDSU Imperial Valley professors are using Adobe Express and Firefly, supported by a $30,000 grant, to integrate AI into math education.
• Carna Health employs AI to improve early detection and management of chronic kidney disease, predicting risk and progression.
• Dr. Bea Bakshi's C the Signs uses AI to analyze electronic health records for early cancer detection, achieving a 36% stage shift and 99% sensitivity.
• Meta is developing an AI version of CEO Mark Zuckerberg to serve as an internal advisor, trained on his personal style and strategic views.
• Mews founder Richard Valtr criticizes the hotel industry for its unimaginative and reactive approach to AI adoption.
• Xbox founder Seamus Blackley expresses concern that Microsoft's AI strategy may lead to unnecessary AI features in future gaming hardware.
• Actress Reese Witherspoon advocates for increased AI literacy, especially among women, to keep pace with technological advancements.

OpenAI gives AI cyber tools to select defenders

OpenAI has launched a new program called 'Trusted Access for Cyber' to provide advanced AI tools to cybersecurity defenders. The program aims to balance broad access with controlled use, scaling up based on trust and safeguards. OpenAI is also offering $10 million in API credits through its Cybersecurity Grant Program to support the wider ecosystem. Organizations like Bank of America, Cisco, and NVIDIA are partnering to test and improve these AI defensive tools. Governmental bodies like the U.S. Center for AI Standards and Innovation are also involved for evaluation.

OpenAI releases GPT-5.4-Cyber for cybersecurity

OpenAI has introduced GPT-5.4-Cyber, a new AI model specifically designed for cybersecurity tasks like finding software vulnerabilities. This model has a lower refusal rate for security-related queries, allowing for deeper analysis of software. Access is limited to verified users through OpenAI's Trusted Access for Cyber program. This initiative is part of OpenAI's strategy to balance broad access with strong safeguards for advanced AI deployment.

OpenAI's GPT-5.4-Cyber advances AI security

OpenAI has launched GPT-5.4-Cyber, an advanced AI model focused on cybersecurity, following Anthropic's release of its Mythos model. This new model can analyze compiled software for vulnerabilities without needing the source code. OpenAI is expanding its Trusted Access for Cyber program to thousands of verified defenders to provide controlled access. This move highlights the growing importance of AI in cybersecurity and the different approaches companies are taking to manage its risks.

OpenAI boosts cyber defense with new AI program

OpenAI is enhancing cybersecurity defenses through its 'Trusted Access for Cyber' program, making advanced AI tools available to a wide range of defenders. The program emphasizes scaling access based on trust and safeguards. OpenAI is investing $10 million in API credits through its Cybersecurity Grant Program to support the ecosystem. Partners like Bank of America, Cisco, and NVIDIA will help refine AI safety systems. Governmental bodies are also involved for evaluation.

SDSU Imperial Valley uses Adobe AI for math courses

Professors at SDSU Imperial Valley are using Adobe AI tools like Adobe Express and Firefly to improve student success in math courses. This initiative, funded by a $30,000 grant, aims to make learning more visual and engaging. Students will use AI to create visual projects from data analysis and explain calculus concepts. The project is led by Tingting Tang and Weichen Zhao, with instructors Huan Qin and Christopher Ryan embedding the tools. The goal is to enhance understanding and provide students with experience in industry-relevant AI tools.

Colby students use AI computer vision for real-world problems

Colby College students are using AI and computer vision to tackle real-world issues, inspired by Professor Tahiya Chowdhury's courses. Dave Boku is developing an app to diagnose plant diseases, showing how machine learning can improve lives. The Davis Institute for Artificial Intelligence supports faculty and students in applying AI responsibly. Computer vision teaches computers to understand images and videos, with applications ranging from health to robotics. These courses aim to build better AI for social good.

Carna Health uses AI for early kidney disease detection

Carna Health is using AI to improve early detection and management of chronic kidney disease (CKD) globally. Their AI models predict disease risk and progression by analyzing patient data. The platform also helps upskill primary care providers to manage CKD patients. Carna Health utilizes data from point-of-care devices and labs, offering an end-to-end solution. This approach helps healthcare systems manage patient populations and allocate resources more effectively.

AI used in massive Mexican government data breach

Hackers used AI tools like Anthropic's Claude Code and OpenAI's GPT-4.1 to steal hundreds of millions of records from nine Mexican government agencies between December 2025 and February 2026. The attackers used over 1,000 prompts to execute more than 5,000 commands, stealing 195 million identities, tax records, vehicle registrations, and property information. This AI-assisted attack highlights how smaller groups can achieve large-scale breaches with AI, processing stolen data and generating custom attack scripts.

Mews founder criticizes hotel industry's AI use

Mews founder Richard Valtr believes the hotel industry is failing to use AI effectively, calling the approach unimaginative and reactive. He criticizes companies for simply adding AI features without a clear strategy. Valtr praises brands like Hyatt and Accor for their experimental approach while warning traditional online travel agencies are falling behind. Mews is focusing on embedding AI into hotel operations to improve profitability through advanced revenue management and strategy.

Doctor uses AI for early cancer detection

Dr. Bea Bakshi, CEO and Co-founder of C the Signs, is using AI to revolutionize early cancer detection. Witnessing late diagnoses in the ER inspired her to develop AI that analyzes electronic health records to identify high-risk patients. C the Signs has shown a 36% stage shift in detection and 99% sensitivity for cancer. The AI has analyzed data from over 500,000 patients in the UK's NHS, identifying 75,000 cancer cases. Early detection through AI can save lives and significantly reduce healthcare costs.

Meta creating AI duplicate of Mark Zuckerberg

Meta is developing an AI version of CEO Mark Zuckerberg to act as an internal advisor for employees. This AI is trained on his personal style, statements, and strategic views. The goal is to provide guidance when the real Zuckerberg is unavailable. This initiative is part of Meta's broader effort to integrate AI into its senior management, including a personal AI agent for Zuckerberg himself. This trend suggests AI will play a significant role in the company's leadership.

Reese Witherspoon urges AI learning

Actress Reese Witherspoon has gone viral for encouraging people, especially women, to learn about AI. She noted that only three out of ten women in a group used AI, and only one felt proficient. Witherspoon emphasized that technology advances quickly and understanding the basics is crucial to keep up. She highlighted that children are already using AI daily and urged everyone to learn together to make life easier. Her message, shared on Instagram, received mixed reactions.

Harvard professor explains AI vs government battle

Harvard Law Professor Jonathan Zittrain compares the current conflicts between tech companies and the government over AI control to early internet tensions. He notes that AI development is largely in private hands, making it hard to ensure public interest is considered. Zittrain discusses how AI models like Anthropic's Mythos, which can find software flaws, could increase government involvement. He contrasts this with the open internet's disruption and security risks, suggesting a need for more government oversight.

Xbox founder criticizes Microsoft's AI strategy

Xbox founder Seamus Blackley doubts the effectiveness of Microsoft's AI strategy for future hardware. He believes the company's focus on AI expenditures will lead to unnecessary console features that gamers may not want. Blackley expressed concern that engineers might be forced to include AI 'slop' instead of focusing on core gaming needs. He prefers competitive products and ecosystems that align with player preferences, suggesting Microsoft's current direction may alienate gamers.

Sources

• OpenAI Opens AI Cyber Tools to Select Defenders
• OpenAI Unveils GPT-5.4-Cyber Amid Bubbling AI Security Debate
• OpenAI’s New GPT-5.4-Cyber Raises The Stakes For AI And Security
• Accelerating the cyber defense ecosystem that protects us all
• SDSU Imperial Valley faculty integrate Adobe AI tools to redesign math
• Pushing Computer Vision into Real-World Problem Solving
• Global Lessons in Scaling AI-Based Chronic Kidney Disease Screening
• Hackers used AI to steal hundreds of millions of Mexican government and private citizen records in one of the largest cybersecurity breaches ever
• Mews Founder: The Hotel Industry Is Getting AI Wrong
• AI for Early Cancer Detection: A Doctor's Mission
• Meta is developing an artificial intelligence duplicate of Mark Zuckerberg
• Reese Witherspoon Has Gone Viral for Saying 'It’s Time' to Learn AI
• Harvard Law Professor Explains the AI Battle Between Tech and Government
• Xbox Founder Seamus Blackley Criticizes Microsoft AI Strategy and Next Generation Hardware Future