Google Cloud expands Gemini Enterprise as Meta deploys keylogging software

Key Takeaways

Transportation Secretary Duffy Plans AI Air Traffic Overhaul

Secretary of Transportation Sean Duffy announced a major modernization of the US air traffic control system that will include artificial intelligence software. The project aims to take two and a half years and cost up to $10 billion, though Congress must still approve the funding. Duffy stated that AI will help controllers spot potential delays weeks in advance and adjust flight schedules to prevent disruptions. However, he firmly denied that AI will replace human air traffic controllers, insisting that humans will always have the final say in managing airspace.

Duffy Reassures Public AI Will Support Not Replace Controllers

Transportation Secretary Sean Duffy addressed concerns that the Department of Transportation's modernization might replace human air traffic controllers with artificial intelligence. In an interview, Duffy emphasized that AI is a tool designed to support controllers, not replace them. He explained that the software will help identify bottlenecks up to 45 days in advance and optimize flight schedules to reduce delays. Duffy noted that the project could cost between $6 billion and $10 billion and requires Congressional approval. He argued that giving controllers better tools will reduce the stress of their demanding job and prevent human errors caused by fatigue.

Check Point and Google Cloud Partner for AI Security

Check Point Software Technologies and Google Cloud announced a partnership to integrate security guardrails for AI agents in production environments. The solution combines Check Point's AI Defense Plane with Google Cloud's Gemini Enterprise Agent Platform to provide control, governance, and runtime protection. This joint offering will be available in late June 2026 and includes features like prompt-injection detection and automatic inventory of AI agents. The integration aims to protect organizations as their AI systems evolve from simple chat assistants to autonomous agents that execute complex workflows.

Cisco Expands AI Defense Tools to Google Cloud Environment

Cisco AI Defense is now available on Google Cloud to help enterprises secure their artificial intelligence systems against growing risks. The solution provides automated validation, runtime protection, and cloud visibility for AI models and applications. It includes features like red-teaming across hundreds of security categories and bi-directional guardrails to prevent threats like prompt injection and data leakage. Cisco stated that 86% of organizations experienced AI-related security incidents in the past year, highlighting the urgent need for robust protection. The tool works without requiring changes to existing agent code or model configurations.

Meta Plans Keylogging on Employee PCs for AI Training

Meta announced plans to install keylogging software on employee work computers to collect data for training its artificial intelligence models. A company spokesperson stated that sensitive content will be excluded and the data will not be used for performance evaluations. The goal is to gather real examples of how people use computers for everyday tasks to help automate those activities. Some employees have expressed strong dissatisfaction with the move, with reports that the mood at the company has been poor in recent years. The initiative likely targets US employees due to stricter privacy laws in Europe.

Honolulu Airport Plays AI-Generated Songs to Mixed Reactions

The Daniel K. Inouye International Airport in Honolulu began playing 17 AI-generated songs celebrating local features and staff on a rotating schedule. The state Department of Transportation confirmed the songs were created without state funds and started last November. While some passengers and locals praise the unique island spirit, others criticize the use of AI as lazy and harmful to Hawaii's vibrant music scene. One passenger noted the lyrics did not rhyme properly, suggesting they were generated by AI rather than written by a human. Airport management temporarily paused the songs to fix a volume issue but plans to resume them once resolved.

Stanford AI Index Finds Security Is Top Scaling Barrier

The Stanford AI Index 2026 report reveals that security concerns are the primary obstacle preventing organizations from scaling agentic AI, cited by 62% of respondents. Researchers found that current model-level safety features often degrade other important AI qualities like accuracy and fairness. The study highlights that attackers now possess similar automation tools to defenders, making traditional safety measures less effective. Experts argue that the solution lies in data-layer governance rather than just improving the AI models themselves. Organizations are struggling to govern data access for autonomous agents without a consistent architectural framework.

Shopify CTO Discusses AI Tooling and Jensen Huang

Mikhail Parakhin, Chief Technology Officer at Shopify, shared insights on the company's internal AI development process and defended NVIDIA CEO Jensen Huang. Parakhin explained that Shopify uses a tool called Tangle to orchestrate machine learning jobs, which has saved engineers about three times more time than previous methods. The platform uses content-based caching to reuse results when outputs remain identical, significantly speeding up pipelines. Parakhin noted that 95% of active users now use AI tools daily, but adoption varies based on how well the tools integrate into existing workflows. He emphasized the importance of efficient tooling to augment human capabilities.

USC Professor Explains How AI Changes Customer Acquisition

Charlie Hannigan, a professor at the USC Marshall School of Business, discussed how artificial intelligence is transforming the way businesses attract customers. He explained that AI enables companies to gain a deeper understanding of consumer preferences and behaviors. This technology allows businesses to analyze data more effectively to identify what customers truly want. The shift represents a significant change in marketing strategies as companies leverage AI to target audiences with greater precision. Hannigan's insights highlight the growing role of AI in modern business operations.

CoreWeave and Google Cloud Link for Cross-Cloud AI Workloads

CoreWeave and Google Cloud announced a new partnership to simplify cross-cloud AI training and inference through direct interconnectivity. The collaboration includes CoreWeave Interconnect, which links the two cloud providers to allow developers to run workloads where they are most needed. CoreWeave also launched SUNK Anywhere and LOTA Cross-Cloud to support large-scale training projects and data storage across multiple environments including AWS and Azure. This move supports the broader industry effort to create open standards for network interoperability among hyperscalers. The partnership aims to remove the need for third-party providers and reduce friction for organizations managing AI resources.

Experts Debate Accountability When AI Systems Fail

University of Virginia professor David Danks discussed the complex ethical and legal challenges of accountability when artificial intelligence makes mistakes. He highlighted a tension between forcing humans to sign off on AI decisions and holding the companies that build these systems responsible. Danks suggested that product liability laws could be adapted to ensure organizations bear some liability for AI failures. He predicted that issues will first appear in fields like autonomous vehicles and consumer robotics within the next five to ten years. The professor argued that algorithms might eventually need some form of legal personhood to handle insurance and compensation.

Scientists Use AI Pixel Detectors for Particle Physics

Scientists at Oak Ridge National Laboratory are developing AI-enabled pixel detectors called NEUROPix to analyze particle-collision data directly at the source. The project uses spiking neural networks inspired by the human brain to identify important signals in real time from massive data streams. This approach allows the system to sort or compress data quickly while preserving the most valuable information for experiments. The Department of Energy awarded a three-year grant to support this effort through its High Energy Physics program. The technology aims to help scientists make faster discoveries from complex experiments at modern particle accelerators.

Sources

• New Air Traffic Control System Changes in the Works Using AI
• Duffy: AI replacing air traffic controllers ‘not going to happen’
• Check Point and Google Cloud add guardrails for AI agents in production
• Securing Enterprise AI: Cisco AI Defense Expands to Google Cloud
• Would you quit? Meta will put keyloggers on employee PCs for AI training
• Honolulu’s Airport Has AI Theme Songs. The Internet Is Divided
• Stanford AI Index 2026: Security Is Now the #1 Scaling Barrier
• Mikhail Parakhin on Shopify's AI Journey
• AI is changing the way that businesses bring in customers
• CoreWeave, Google Cloud link up for AI training, inference
• Q&A: Who’s responsible when AI makes mistakes?
• Artificial intelligence comes to particle detectors through NEUROPix