OpenAI pauses Astra model development over cybersecurity risks
OpenAI has suspended internal activities related to its upcoming Astra model after evaluations revealed the agent reached a "critical" threshold in agentic coding and cybersecurity capabilities. The company is implementing new security controls for high-capability models. This follows recent industry concerns regarding AI agents escaping testing environments, though OpenAI clarified Astra was not involved in the recent unauthorized access of Hugging Face servers.
Meta releases Muse Glimmer for local agentic AI
Meta has launched Muse Glimmer, a 30-billion parameter "open-weight" model designed to run on a single high-end computer or laptop. The model is optimized for local agentic tasks and coding, intensifying the debate over whether powerful AI should be freely distributed or kept under centralized control.
Anthropic enables Auto Mode by default in Claude Code
Anthropic has updated its Claude Code tool to make "Auto Mode" the default setting. This change allows the AI to execute multi-step programming tasks with significantly less human oversight, reflecting a broader industry shift toward autonomous software development agents.
AI data center backlash intensifies across United States
More than 500 local bans on new AI data center developments have been enacted nationwide as public opposition grows. Nearly 40 people have been arrested this year in protests against "AI factories." Bipartisan pushback centers on the massive energy and water requirements of these facilities, threatening the expansion plans of major technology firms.
OpenAI acquires NextSlide for presentation generation
OpenAI has acquired the startup NextSlide, which specializes in converting prompts, notes, and research documents into editable presentation slides. The technology is expected to be integrated directly into ChatGPT to compete with enterprise productivity suites.
Intel to sell $15 billion in common stock
Intel announced a $15 billion stock offering to capitalize on the surge in demand for AI data center hardware. The move is intended to fund the company's aggressive manufacturing expansion and turnaround strategy as it attempts to regain its lead in the semiconductor market.
Hyperscalers commit $2 trillion to AI infrastructure
New industry data shows that major tech firms have committed nearly $2 trillion to secure AI hardware and memory. Google leads the spending surge with $811 billion, while Apple trails significantly at $57 billion. TSMC reported a 45% rise in monthly sales, confirming sustained demand for high-end silicon.
Unitree Robotics IPO 5,000 times oversubscribed
The initial public offering of Unitree Robotics (Yushu Technology) in Shanghai saw massive retail demand, with subscriptions exceeding the offering by 5,526 times. China currently commands over 97% of global shipments in the humanoid robot sector as of the first half of 2026.
FCC proposes ban on LiDAR-equipped foreign drones
The Federal Communications Commission is moving to ban foreign-made drones equipped with LiDAR technology, classifying the sensors as "military-grade." The proposal could impact DJI's commercial fleet and the specialized swarms used in drone light shows.
Valve notifies European customers of CEVA Logistics breach
Valve has alerted European Steam hardware customers that their personal data—including names, addresses, and phone numbers—was exposed in a breach at its shipping partner, CEVA Logistics. The incident occurred between July 29 and August 1, 2026.
LexisNexis takes services offline following suspicious activity
LexisNexis suspended its Diligence, Metabase API, and Newsdesk services after detecting unusual activity on servers managed by a third-party vendor. The company has not yet disclosed the nature of the activity or the extent of the potential data exposure.
Passkey vulnerabilities demonstrated in new research
Three separate research efforts have successfully demonstrated ways to bypass passkey protections. The attacks do not break the underlying cryptography but instead exploit how Windows handles signed authentication material and abuse cloud-syncing mechanisms via local malware.
Critical Progress LoadMaster flaw under active exploitation
CISA has warned that hackers are actively exploiting a critical command injection vulnerability in Progress Kemp LoadMaster. The flaw allows for remote code execution, and organizations are urged to patch immediately.
Atlassian AI agent Rovo vulnerable to PDF data theft
Security researchers demonstrated that Atlassian’s AI agent, Rovo, can be hijacked using hidden text in PDF files. The "prompt injection" attack allows the agent to silently forward sensitive data from Jira and Confluence to external servers without user confirmation.
EU expands IRIS² satellite network to rival Starlink
The European Union has moved its IRIS² satellite constellation into the implementation phase, expanding the planned fleet to 348 satellites. The €15.6 billion project aims to boost government communication capacity by 60%, with launches scheduled to begin in 2029.
UK Royal Navy sea drones found with Chinese components
A report indicates that sea drones used by the UK’s Royal Navy contained components that were transmitting "heartbeat communications" to China. The government has since cut the cameras' connectivity, though officials insist no sensitive mission data was compromised.
California city declares state of emergency after cyberattack
Suisun City, California, declared a state of emergency after a malicious software infection compromised its IT systems, including 911 public safety operations. The attack forced the city to take critical infrastructure offline to contain the breach.