Latest Software & Apps News
Anthropic Launches Project Glasswing to Secure Critical Software with AI
Anthropic announced Project Glasswing, uniting AWS, Apple, Google, Microsoft, and others to leverage the Claude Mythos Preview model for finding vulnerabilities in OS and browsers. The initiative commits $100M in credits and $4M donations to bolster defensive cybersecurity against AI-driven threats. Over 40 organizations gain access to scan critical infrastructure.
DeepSeek V4 Model Set to Test China's AI Ambitions on Huawei Chips
DeepSeek's upcoming V4 multimodal AI model, potentially trained on Huawei chips, aims to bypass US Nvidia restrictions, with involvement from Alibaba, ByteDance, and Tencent. Seen as a benchmark for China's AI progress, it could shock US tech valuations with low-cost, high-capability features. V4 follows V3 from December 2024 and targets massive context windows.
FBI Warns of Chinese Mobile Apps Exposing Sensitive User Data
The FBI issued an alert about certain Chinese mobile apps that could expose sensitive user data to risks. Officials highlight vulnerabilities in apps available for download. Users are advised to exercise caution with such applications.
Microsoft VP: AI Layoffs Will Boost Software Companies' Business
Microsoft VP Rajesh Jha stated that even if companies lay off half their workforce due to AI, it will increase business for software firms rather than harm them. He emphasized AI's role in enhancing productivity and demand. This counters fears of AI 'killing' software jobs.
AI Agents Raise Liability Questions in Business Automation
Discussions emerge on AI agents promising to 'run the business,' but liability issues arise if errors occur. Experts debate accountability in autonomous software systems. This highlights growing concerns in enterprise AI adoption.
New Cryptography Trends Shift from Algorithms to Patentable Systems
Recent trends in cryptography focus on integrated systems rather than isolated algorithms, opening avenues for patentable innovations in software security. Published insights detail evolving practices. This could impact app and software development standards.
Claude Mythos Preview Finds Vulnerabilities in Every Major OS and Browser
Anthropic's unreleased Claude Mythos Preview model has identified thousands of high-severity vulnerabilities across all major operating systems and web browsers. It outperforms prior models on security benchmarks like CTI-REALM. This underscores AI's transformative potential in cybersecurity.
Project Glasswing Partners Commit to Defensive AI Security Use
Launch partners including Cisco, CrowdStrike, NVIDIA, and JPMorganChase will use Mythos Preview for defensive security. Anthropic shares learnings industry-wide to counter proliferating AI exploit capabilities. The effort aims to give defenders an advantage.
Alibaba Opens AI Data Center with 10,000 Domestic Chips
Alibaba launched a new AI training data center in southern China using 10,000 of its own chips, operated by China Telecom. This supports models like DeepSeek V4 amid chip restrictions. It signals China's push for AI self-reliance.
DeepSeek V4 Expected as Open-Source Multimodal Powerhouse
Analysts predict DeepSeek V4 will be a highly capable open-source model generating text, images, and video at low cost. It builds on V3's popularity in China and emerging markets. Success could reshape global AI competition.
Anthropic Commits $104M to Open-Source Security via Glasswing
Anthropic pledges up to $100M in Mythos Preview credits and $4M in donations to open-source security groups as part of Project Glasswing. Extended access goes to 40+ critical software maintainers. This accelerates industry-wide defenses.
China's Tech Giants Stockpile Huawei Chips for DeepSeek Launch
Alibaba, ByteDance, and Tencent have ordered large quantities of Huawei's latest chips for the DeepSeek V4 rollout. This preparation challenges US export controls on advanced AI hardware. It marks a potential geopolitical shift.