Latest Software & Apps News
Anthropic Announces Project Glasswing to Defend Against AI-Powered Vulnerabilities
Anthropic unveiled Project Glasswing, a major cybersecurity initiative bringing together Amazon Web Services, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorganChase, Linux Foundation, Microsoft, NVIDIA, and Palo Alto Networks. The project leverages Claude Mythos Preview, an unreleased AI model that has found thousands of high-severity vulnerabilities in every major operating system and web browser, to help defenders secure critical software before malicious actors weaponize such capabilities
.
AI Models Surpass Human Experts in Finding Software Vulnerabilities
AI coding models have advanced to the point where they can exceed all but the most skilled humans at identifying and exploiting software flaws, according to developers maintaining critical cyber infrastructure. This capability emerged noticeably in early 2026 following releases of cutting-edge models in late 2025
, shifting the cybersecurity landscape from theoretical AI threats to practical vulnerabilities discovered in widely-used systems.
Mythos Preview Model Restricted to 50 Organizations for Security Reasons
Anthropic is limiting access to Mythos Preview to approximately 50 select companies and organizations as part of Project Glasswing, with no plans for general public release due to misuse risks. The company committed up to $100M in usage credits and $4M in direct donations to open-source security organizations to support defensive security work
.
Global AI Adoption Reaches 88% Among Organizations
Generative AI adoption has surged to 88% of organizations using it in at least one core business function as of April 10, 2026, up from 71% in 2025. The global Generative AI market has reached $161 billion in 2026 and is projected to reach $1.2 trillion by 2034
.
Enterprise Transition from Chatbots to Agentic AI Systems
Organizations are shifting from simple question-answer interfaces to autonomous agentic AI systems capable of searching databases, cross-referencing market data, and executing complex transactions without human supervision. The week of April 4-10, 2026 marked a critical transition point, with enterprise-level adoption declarations and new product introductions accelerating as companies dedicate greater resources to full operational AI implementation rather than small-scale pilots
.
Salesforce Reports 30% Productivity Gains with AI Coding Agents
Salesforce CEO Marc Benioff announced that the company's 15,000 engineers have achieved over 30% productivity increases by working alongside AI coding agents from Anthropic, OpenAI Codex, and Cursor. Salesforce froze engineer hiring throughout fiscal year 2026, using AI agents to maintain workload capacity while adding approximately 20% more sales staff to meet market demand
.
AI Handles 30-50% of Salesforce Workload with 93% Accuracy
Salesforce reports that AI now accounts for 30% to 50% of the company's total workload at approximately 93% accuracy, representing what CEO Marc Benioff characterizes as a 'digital labor revolution'. Despite productivity gains, Salesforce continues hiring engineers at scale, indicating that human oversight remains essential as AI models cannot yet operate fully autonomously
.
Linux Kernel Maintainers See Vulnerability Detection Improvements
Jim Zemlin, CEO of the Linux Foundation which maintains the Linux kernel powering Android and the world's 500 most powerful supercomputers, confirmed that Mythos Preview may provide better capabilities for helping developers fix vulnerabilities than previous AI models. The Linux kernel's critical infrastructure role makes improved vulnerability detection capabilities particularly important for global software security
.
HackerOne Develops Autonomous AI for Vulnerability Detection and Remediation
Security firm HackerOne is developing an agentic AI product capable of more autonomously finding and fixing vulnerabilities, representing a shift toward fully automated security threat response. This development reflects the growing sophistication and autonomy of AI systems in cybersecurity operations.
2026 Marks Widespread Technology Adoption Year Following Exploration and Testing Phases
Industry analysts characterize 2026 as the beginning of widespread AI technology adoption, following 2023 as an exploration period and 2024 as a testing phase. Enterprise organizations are now transitioning from research pilots to operational deployment across entire business functions, marking a fundamental shift in how organizations integrate generative AI into their core operations
.