FBI Flags Cyber Incident Involving Surveillance Systems 

FBI Cyber relentlessly targets cyber attackers by leveraging unique authorities, advanced capabilities, and strong partnerships.  

Hijacked networks, cryptocurrency theft, and corporate espionage highlight the growing cyber threat. Each year, adversaries grow more skilled, attacking energy networks, shutting down hospitals, and heightening global tensions. Government-backed actors target the US and critical infrastructure, while cyber criminals exploit vulnerabilities to steal money and extort data.  

Combating these threats is the FBI’s cyber program’s primary mission. As the lead federal agency for investigating cyber attacks and intrusions, we engage with victims and work to unmask those committing malicious cyber activities wherever they are.  

Learn more about nation‑state cyber attacks. Report cybercrime by visiting our official website and following the bureau’s ongoing efforts to combat the growing cyberthreat.  

FBI Cyber on LinkedIn and X.  

A Larger Than Ever Attack Surface. 

Americans have witnessed major digital advances from smart cities to self-driving cars and AI technology. This expanding digital world also gives hostile actors more opportunities to attack. As we adopt these technologies, we must confront rising threats from both nation-states and cybercriminals.  

China, Russia, Iran, and North Korea continue cyber attacks against US targets. Some grow more aggressive, infiltrating critical infrastructure and private networks. They seek not just intelligence and intellectual property but also the ability to harm the US.  

Ransomware attacks can shut down companies and disrupt industries. Attackers operate in a complex network of developers, partners, and service providers. Purdes consistently evolving methods  

Sharing Actionable Intelligence 

The FBI connects with public and private sectors, building strong relationships and sharing useful threat information with key stakeholders.  

Sharing information is a key part of our mission. We use our authority to warn the public about new online threats and to advise industry partners on how to find and stop them. We send out alerts as quickly as possible because threats and attack methods change fast.  

How FBI Cyber Works 

FBI Cyber’s scale team spans the US and 20 countries, ready to help after cyber incidents. Our mission: serve victims by sharing information and tools to address threats in real time.  

We use our partnerships, expertise, global reach, and unique investigative and intelligence authorities to help victims in every engagement.  

  • The FBI has specially trained cyber squads in each of our 56 field offices, working hand in hand with interagency task force partners.  
  • The FBI leads the National Cyber Investigative Joint Task Force (NCIJTF), uniting over 30 D agencies from intelligence and law enforcement.  
  • The Rapid Response Cyber Action Team can deploy across the country within hours to respond to major incidents.  
  • Cyber assistant law enforcement attaches in the US and worldwide enable close work with international partners to seek justice for cybercrime. Victims  
  • The Internet Crime Complaint Center (IC3) collects reports of internet crime from the public. Using such complaints, the IC3’s recovery asset team has helped freeze hundreds of thousands of dollars for victims of cybercrime.  
  • CYWatch is the FBI’s 24-7 operations center and watch flow, providing around-the-clock support for tracking incidents and communication.  

The FBI has declared a recent data breach of a surveillance system a major incident and notified Congress, the agency told The Hill on Friday.  

FBI identified unusual activity on an unclassified network and quickly used its technical capabilities to address the incident. Access was obtained via a third party and classified as a major incident under the Federal Information Security Modernization Act (FISMA).  

The FBI is following the required steps under FISMA, including notifying Congress, and remains focused on countering nationwide nation-state and cybercriminal activity. It continued.  

Several news outlets have speculated that hackers linked to China may have carried out the breach.  

Politico reports that the system held data from pen register and trap-and-trace devices, which track incoming and outgoing phone calls. The breach affected the personally identifiable information of individuals involved in FBI investigations, highlighting the risk of exposure for them.  

The government defines a major incident under FISMA as any incident likely to cause demonstrable harm to the national security interests, foreign relations, or economy of the United States, according to 2025 guidance. An incident involving personally identifiable information may also be classified as major if it poses risks similar to those in this breach. 

Source:The Cyber Threat 

Meta Shifts Messaging Toward Integrated AI Platforms 

Meta has changed its strategy, moving away from separate social networking apps and focusing on building deeply integrated, automated communication platforms. The change, which accelerated in early 2026, aims to connect WhatsApp, Instagram, and Messenger using advanced reasoning engines. Meta wants to turn messaging threads into workspaces where users can handle complex tasks. The goal is to make digital experiences smoother so people can easily move from chatting to actions like booking, travel, managing projects, or organizing shared information all in one place.  

The Architecture Of The Interactive Workspace 

At the core of this change is the persistent context framework, which lets automated assistance track user preferences across apps. Previously, digital assistants forgot details after a task. Now the Matters platform uses a memory system, so a conversation started on Instagram can continue WhatsApp. For example, if a user mentions a hiking trip in a group chat, the system can later show trail maps, weather updates, and gear lists in a private message.  

This integration is powered by new interoperable protocols that enable messaging apps to directly connect with select third-party services. Standard APIs allow assistants to check flights or book restaurants without extra apps, making messaging the main hub for digital life management.  

Improving Business Logic and Customer Relations 

Meta’s integrated platforms are changing how conversational commerce operates for businesses. The new business intelligence tier on WhatsApp and Messenger targets small and medium businesses, letting merchants use automated agents to manage the entire customer journey—from product discovery and recommendations to payments and post-sale support—by analyzing prior interactions and purchases. The system delivers personalized service similar to that of a dedicated human assistant.  

A brand consistency system ensures automated replies reflect the company’s tone, language, and policies for industries like airlines or global retail. This delivers 24/7 multilingual support without extra staffing. Complex or sensitive queries are sent to human agents along with the full chat history and an action plan from the reasoning engine.  

Privacy Safeguards And Data Sovereignty 

As automated systems become more common in private conversations, data security has become a top priority for Meta. The company uses a secure processing model with trusted execution environments (TEEs) to process sensitive tasks. In this setup, data for a specific request is processed in a secure hardware area that the rest of the system cannot access. After the task is finished, such as verifying a payment or retrieving a private calendar entry, the temporary data is deleted from memory. This ensures that personal information is not stored or used for any purpose other than the one for which it was collected.  

Meta has also updated its end-to-end encryption (E2EE) to cover interactions between users and automated agents. This means that the details of digital assistance tasks are protected by the same security as private messages. Users gain an extra layer of digital sovereignty, enabling them to use automation without sacrificing privacy. The system includes a transparency dashboard that lets users view and manage the data permissions they have granted to different automated services, giving them greater control over their digital information.  

The Evolution Of Social Connectivity 

Meta’s long-term plan is to build a social knowledge graph that shows the detailed relationships and shared interests of people around the world. Building on this, Meta aims to offer a search and discovery experience that feels more personal than traditional keyword searches by leveraging the knowledge of its many users. For example, if someone needs a recommendation for a professional service, the system can show results from friends or contacts who have reviewed or used it. As a result, the messaging platform becomes a trust-based navigator where recommendations from peers matter more than regular ads.  

The Quiet Resonance of the Digital Thread 

As digital technology advances, online interactions steadily evolve. Digital tools are becoming more responsive and helpful, integrating seamlessly into daily life. The era ahead merges messaging and action, creating a unified online experience. While the line between human and machine input may blur, individuals will continue to rely on technology while maintaining control and security.  

Source: https://about.fb.com/news/ 

Amazon Patent Shows Gesture-Based Control for Devices

Amazon holds several patents for gesture-based control systems in devices such as drones, wearables, and home entertainment products. Technologies intend to make it easier to interact with machines using simple, touch-free movements.  

Below are key details about Amazon’s gesture-control patents.  

  • Gesture-recognizing drones: Amazon’s drone system patent could make drone delivery easier by allowing users to interact with drones via gestures and voice commands. The drones’ ability to respond to human movement may help make deliveries safer and more user-friendly.  
  • Warehouse wristbands: The ultrasonic wristband patent aims to increase operational efficiency in Amazon warehouses by improving picking speed and accuracy. This patent is relevant to optimizing inventory management, thereby lowering costs and improving overall warehouse productivity.  
  • Child-appropriate interface selection: This patent helps ensure user safety by automatically adapting device interfaces for younger users based on hand size. It is relevant to digital well-being, providing age-appropriate experiences that protect and simplify children’s technology use.  
  • Media device control: This patent allows users to operate media devices using hand gestures for a more intuitive, remote-free experience. It is relevant as it could enhance accessibility and user interaction by reducing dependency on physical remotes.  

These patents, reflecting Amazon’s innovation, were developed primarily by its Lab 126 division.  

Building on its previous advancements, Amazon first impressed millions with Alexa’s intuitive capabilities. Now, the company’s latest project could let drones respond to human hand signals. Recently, Amazon was granted a new patent by the US Patent and Trademark Office (USPTO) for aid delivery.  

Amazon aims to develop a fleet of unmanned aerial vehicles to deliver packages to customers in 30 minutes or less. The patent specifies that the drone can change its behavior in response to various gestures, such as a thumbs-up, shouting, or waving arms. According to the patent, the drone can release its package, alter its course to avoid obstacles, ask people questions, or cancel the delivery.  

Giving Commands To The Machine 

The patent includes several illustrations. In one, a person stands outside a home, waving their arms to show how someone might signal a drone. Overhead, a voice bubble from the man’s mouth suggests he is giving commands to the drone.  

The human recipient and/or other humans can communicate with the vehicle using hand gestures to guide it along its path to the delivery location. The patent says  

Another diagram shows the steps the drone would follow to read human body language during deliveries. Receive a human gesture, access the gesture database, and determine the human gesture based on the gesture database. Proceed in accordance with the determined human gesture and delivery instructions.  

The drones’ communication system would use several sensors, including a depth sensor and cameras that detect visible, infrared, and ultraviolet light. The drones could recognize hand and body gestures, human voices, and movement, such as someone walking toward or away from the drone.  

Other Devices Using Gestures to Control 

Specifically, according to the patent, if drones are authorized to deliver, they can either drop packages with extra padding from the air or land and hand them over.  

Since announcing its plans for air delivery, Amazon has filed several patent applications. These include giant airships that could act as flying warehouses and drones that self-destruct in emergencies. Amazon has not said when, or even if, the gesture-recognition drones will be available.  

For example, Amazon made its first autonomous drone delivery to a customer in the United Kingdom in 2016. Additionally, the company is currently conducting a private customer training trial for drone delivery in Britain.  

Source: Latest Amazon Patent Includes Gesture-Recognizing Drones 

Apple Expands Business Tools to Challenge POS Systems

Retailers everywhere are quickly swapping out old counter POS systems for portable mobile-based sales tools thanks to advances in 5G, eSIM technology, and retail software. Everyday devices like phones and tablets now serve as fully functional POS systems.  

Earlier, we explored the popularity of tablet POS. Nowadays, many retailers use smartphones for sales.  

The iPhone is now among the most widely used POS systems, supported by many iOS sales apps. The mobile POS (mPOS) industry is changing in-person retail from stalls to chains.  

Email receipts are replacing printed ones, and contactless payments now surpass cash. Likewise, mPOS systems show how wireless technology is changing shopping.  

The Rise Of Mobile Point Of Sale (mPOS) 

The 2018 iPhone XR enabled Apple Tap to Pay, allowing iPhones to function as full POS terminals for contactless payments.  

Handheld POS systems, such as iPhones, offer numerous opportunities. However, these new formats introduce challenges that retailers must address to fully benefit from mPOS.  

What To Look For In A New POS System 

Whether new or established retailers should select an APOS system based on store type, desired customer experience, and location. Four key factors warrant contribution:  

  • Modern POS software manages sales, receipts, inventory, and staff timesheets in a single package. iPhones use many retail apps from the App Store, like Shopify and Square, and can add more for specialized needs. Some require external card readers; others work directly on the iPhone.  
  • Customer experience: the POS system is often the last point of contact with customers, so it strongly shapes their impression of your store and service. Apple POS systems are familiar and seen as a gold standard in consumer tech, which helps build trust during transactions. Because iPhones and iPads are portable, staff can complete sales anywhere in the store, reducing queuing time and making the experience more personal than with fixed terminals.  
  • Data migration is a major challenge when upgrading POS systems, especially preserving old transaction records. Cloud solutions like iOS POS prepare retailers for the future and ensure data transfer, cloud storage, and iCloud backups make switching easy. Providers like 1Global use remote SIM provisioning to quickly set up new iPhone POS systems, enabling teams to work efficiently.  
  • Connectivity: Every POS system needs a reliable, fast, and secure internet connection. Most fixed terminals with internet have replaced wireless devices using eSIMs. These devices, whether dedicated POS systems or iPhones, can connect to the internet and work anywhere with mobile data as more stores use eSIMs. In-store mobile connectivity has become a sensible option, especially with retail-focused services like 1Global SafeRetail.  

Why Retailers Are Turning to Mobile POS 

  • Rapid Market Entry And Expansion  

Many new retailers are surprised that consumer technology now matches the capabilities of dedicated point-of-sale hardware. This phenomenon is sometimes referred to as “disbelief mPoster syndrome”.  

More retailers are adopting mobile POS, newer iPhone models, and fast wireless connections now offer the same features as traditional POS systems, all in a compact device.  

Established brands with existing POS systems benefit from the low cost and quick setup of widely available iOS devices. According to WooCommerce, businesses can set up Tap to Pay on an iPhone in just two minutes.  

The flexibility helps brands expand into new markets. It also provides a backup if devices are lost or damaged. The rise of pop-up shops and short-term retail is another benefit, as brands can test new formats and reach different customers without investing in permanent stores. Pop-ups help brands enter new markets and capitalize on seasonal events such as festivals, conventions, and trade fairs.  

  • Scalability  

Retail is competitive. Brands need POS solutions that grow with them, using cloud-connected devices such as iPads or iPhones, and reduce IT and accounting costs. This makes it easier to expand and stay connected with regular updates.  

For multinational chains, MPOS can use a simple global provider, such as 1Global, to support services worldwide. This eliminates complex contracts in each country and speeds device deployment. The resulting savings let brands offer better consumer prices.  

IOS is familiar to developers. There are now hundreds of POS apps and services for retailers, many tailored to specific needs. Shopify and Zettle let merchants manage offline and online sales in a single app, reducing the need for extra hardware.  

This is in contrast to the ring‑fenced world of fixed-line systems; unlike traditional POS systems, which are limited to a few companies, the open iOS platform enables retailers to create custom mPOS apps tailored to their exact needs, including on iPads. Retailers must purchase a separate NFC reader.  

iPhone POS systems lower costs and reduce hardware needs. Any iPhone XR or newer can act as an NFC card reader without extra devices.  

  • Familiar UI and Faster Onboarding  

Training staff on new POS systems takes time, especially for short-term seasonal roles. A familiar interface like iOS helps reduce building errors and shorten training time. iOS is the most popular smartphone OS in the USA. In self-checkout stores, a familiar user experience is very valuable.  

Apple Tap to Pay 

A key difference between iPad and iPhone POS systems is Apple Tap2Pay. With this feature, iPhones can accept contactless payments from cards or mobile wallets like Apple Pay, Samsung Pay, or Google Pay. No extra hardware is needed. iPads still require an external card reader for payments.  

Tap2Pay works anywhere with a mobile internet connection when paired with a service like 1Global Safe Retail iPhone POS. Let’s vendors process transactions almost anywhere while staying secure. Tablet POS terminals already have lower setup costs than traditional systems, and mPOS options cut costs even further by eliminating the need for additional hardware.  

Is an iPhone POS Secure? 

Can a personal phone really match the security of dedicated sales hardware? Apple Tap to Pay offers the same merchant security as a traditional POS system. No card details are stored on the device. All transactions happen within Apple‑approved POS apps. They also meet the latest PCI standards.   

Nonetheless, vendors and clients take additional steps to safeguard their own security and their customers’ information when using iPhones. The convenience of mPOS systems can make them a potential target for the future.  

Another key step is to use a secure wireless data connection. This is especially important when processing transactions on the move. Avoid unknown Wi‑Fi networks. When traveling or running a temporary stall, use a secure mobile data service like Safe Retail from 1Global to keep all data safe and reliable.  

How Mobile POS Improves the Customer Experience 

Budget reduction efforts can sometimes hurt the customer experience. For example, while automated chatbots may save brands money at first, they can degrade customer care quality and erode trust over time.  

MPOS systems stand out for lowering merchants’ costs and improving the customer experience. Long wait times hurt sales and drive customers away. mPos systems help in two main ways.  

  • Portability: employees can carry a pocket-sized iPhone and process transactions anywhere in the store. This removes the need for a fixed checkout, avoids bottlenecks during busy times, and lets staff offer a more personal experience, including payment.  

iPhone and iPadOS systems with eSIM use mobile data, not just Wi‑Fi. This allows them to work anywhere with a mobile signal and lowers the chance of outages. Services like Safe Retail and 1Global automatically reconnect devices to the next network for smooth sales.  

While mPOS still faces some challenges, the future looks bright for iOS retail. The latest iOS update, iOS 26 Real, reaffirms Apple’s commitment to eSIM connectivity by introducing several new eSIM features.  

iPhone can’t yet process swipe cards, gift cards, or cash, nor print physical receipts. Still, as shopping becomes more digital, mPOS is expected to keep growing from small vendors to global chains. Handheld POS systems open new business opportunities and accelerate expansion. Possessing a reliable network provider like 1Global is key to building a sustainable retail model.  

Source: How iPhones and Mobile Point of Sale systems are reshaping retail 

Google Cuts a Memory Cost With New TurboQuant Method

Vectors are the basic tools AI models use to process information. Simple vectors describe points while high-dimensional embeddings capture complex details, dash-like features of an embedding image, or meanings of words but use much more memory. This can slow down the key value cache, A Fast storage for frequently accessed data, so the computer has to search a slower database.  

Vector quantization is a classic data compression method that reduces the size of multidimensional vectors. This helps AI in two ways. It speeds up vector search, which is the first technology behind large-scale AI and search engines, and it reduces key‑value cache slowdowns by making key‑value pairs smaller. This means faster searches and lower memory cost. However, traditional vector quantization often incurs a memory cost because most methods must compute and store precise quantization constants for each small block of data. This can add 1 or 2 extra bits per number, partly defeating the purpose of compressing the data.  

Today, we are introducing TurboQuant, a new compression algorithm that addresses the memory load value. In vector quantization, TurboQuant uses two other methods: quantized Johnson Lindenstrauss (QJL) and polar quant to achieve its results. In our test, all three techniques help reduce key-value bottlenecks without lowering AI model effectiveness. This could have a big impact on any use case that relies on compression, especially in search and AI.  

How TurboQuant Works 

TurboQuant is a compression method that significantly reduces model size without sacrificing accuracy. It is suitable for both key value compression and vector search. Its approach involves two main steps. Each builds on concepts introduced by Polar Font and QJL.  

  1. High-value compression (the polar quant method): Turboquant starts by randomly rotating data vectors. This clever step simplifies the data geometry, making it easy to apply a standard, high-quality quantizer (a tool that maps a large set of continuous variables, such as precise decimals, to a smaller discrete set of symbols or numbers, such as integers) to each part of the vector individually. This first step uses most of the compression power (the majority of the bits) to capture the main concept and strength of the source vector.  
  1. Next, TurboQuant applies QJL to the remaining error using a single bit, thereby improving the accuracy of the attention score.  

To better understand TurboQuant’s efficiency, let’s examine the specific roles that QJL (quantized Johnson Lindenstrauss) and Polar Quant (polar coordinate quantization) play in its two-step process.  

QJL The Zero Overhead 1-Bit Trick  

QJL uses the Johnson–Lindenstrauss transform, a mathematical method, to compress high-dimensional datasets while preserving important distance relationships between data points. It converts each vector element to a single sign bit, either +1 or −1. This creates a fast shorthand that requires no extra memory. To keep the results accurate, QJL uses a special estimator that bridges a high-precision query with a simpler, lower-precision dataset. This helps the model accurately calculate the attention score, which determines which parts of its input matter most and which can be ignored.  

PolarQuant: A New Angle Of Compression 

Polar Quant solves the memory cost problem in a different way. Instead of using standard coordinates like x, y, and z to represent distance along each axis, Polar Quant converts the vector to polar coordinates. This is like saying go five blocks at a 37-degree angle instead of going three blocks east and four blocks north. This gives two pieces of information. The radius shows the strength of the data, and the angle shows its direction and/or meaning because the angles follow a known pattern. The model does not need to perform a costly data normalization step. It maps data onto a fixed, predictable circular grid with predefined boundaries, rather than a square grid whose boundaries keep changing. This lets Polar Quant avoid the memory overhead of older models.  

Experiments And Results 

We tested all three algorithms on standard long-context benchmarks, including Long Bench, Needle in a Haystack, Zero Scrolls, RULER, and L-Eval, using open-source LLMs such as GAMA and Mistral. TurboQuant delivers top performance across both dot-product distortion and recall while using less key-value (KV) memory. The chart below summarizes how TurboQuant, PolarQuant, and KIVI baselines performed on tasks such as question answering, code generation, and summarization.  

The chart below shows how the algorithms performed on long-context needled-in-hashtag tasks, which test a model’s ability to find information hidden in large text. TurboQuant achieved perfect results across all benchmarks and reduced the key-value memory footprint by at least 6x. PolarQuant performed almost as well for this task.  

TurboQuant can reduce the key-value cache to just 3 bits without any training or fine-tuning, while maintaining model effectiveness. It also runs faster than the first LLMs (Gemma and Mistral). TurboQuant is easy to implement and adds almost no extra runtime. The plot below shows that 4-bit TurboQuant A can be up to 8× faster than 32-bit unquantized keys on H100 GPUs.  

This makes it ideal for supporting use cases. This makes TurboPoint a great fit for jobs like vector search, where it can speed up index building. We tested TurboQuant with high-dimensional vector search against top methods like PQ and Rabbi Q using 1@K recall ratio. This ratio shows how often the algorithm correctly identifies the true top inner product result among its top K guesses. TurboQuant consistently achieved higher recall ratios than the baseline techniques, even though the baselines use large codebooks and require tuning for each dataset. This shows that TurboQuant is both strong and efficient for large-scale search shifts in high-dimensional search, creating a new benchmark for achievable skill. It delivers near-optimal distortion rates in a data-oblivious manner. This shows that our nearest-neighbor engines operate with the efficiency of a 3-bit system while maintaining the accuracy of much heavier models.  

Peering Forward 

TurboQuant, QJL, and PolarQuant are not simply practical engineering solutions. They are also important algorithmic advances supported by strong logical proofs. These methods work well in practice and are often proven efficient, operating close to the theoretical maximum. This solid foundation makes them reliable for large, critical systems.  

One main use of these methods is to solve key‑value cache bottlenecks in models like Gemini, but efficient online vector quantization has an even wider impact. For example, modern search is moving beyond keywords to understand intent and meaning. This shift needs vector search, which finds the most relevant items in a database with billions of vectors. Techniques like TurboQuant are essential. They enable the construction and querying of large vector indexes with minimal memory usage. Almost no pre-processing, along with top accuracy, makes semantic search at scale faster and more efficient. This is important as AI is integrated into more products. Advances in vector quantization will become even more important.  

Source: TurboQuant: Redefining AI efficiency with extreme compression 

Microsoft 365 E7 Adds Control Layer for Enterprise AI Agents 

Today, we introduced the next step in bringing frontier transformation to life for customers in every industry with Wave 3 of Microsoft 365 Copilot, Microsoft Agent 365, and Microsoft 365 E: the Frontier Suite.  

As more customers use agentic AI, CIOs, CISOs, and security teams have important questions. How can I track and monitor these agents? How do I know what they are doing? Do they have the right access? Can they leak sensitive data? Are they safe from cyber threats? How do I manage those threats?  

These new solutions mark a significant advancement in providing clarity and security for organizations adopting AI with Agent 365 and Microsoft 365 E7, the Frontier suite available starting May 1, 2026. We address these questions to give our organizations the confidence to use AI more fully.  

Agent 365: The Control Center for Agents 

As organizations use more agentic AI, gaps in visibility and security can make it easier for agents to act against company interests. Without a control center, teams cannot see which agents exist, their behavior, access, or risks. Microsoft Agent 365 gives you a unified control center, so IT, security, and business teams can observe, manage, and secure agents across your organization whether built on Microsoft AI platforms or from partners using new Microsoft security features that integrate into your workflows.  

Here’s how this works in real situations.  

Now that Agent 365 is running in production, Avanade can clearly see agent activity, manage agent activity growth, control resource use, and treat agents as identity-aware digital entities in Microsoft Intra. This greatly reduces operational and security risks, represents a major step toward overseeing agents at scale, and demonstrates Microsoft’s commitment to responsible production-ready AI. Aaron Reich, Chief Technology and Information Officer, Avanade.  

Key Features Of Agent 365 Include: 

Visibility For Every Role. 

With Agent 365, IT, security, and business teams can view all managed agents, understand their usage, and act quickly on relevant performance, behavior, and risk signals within their current workflows.  

  • The agent registry lists all agents (AI systems that perform your tasks) in your organization, including those built with Microsoft AI Power Partners and those added via APIs (software interfaces that allow product programs to interact).IT teams can access this list in the Microsoft 365 admin center. Security teams can see the same list in their Microsoft Defender, and Purview works.  
  • Agent behavior and performance tracking provide reports on agent performance, usage of metrics, maps, and activity.  
  • Agent risk signals in Microsoft Defender, Intra, and Purview help security teams assess agent risk  as they do for users by detecting issues such as compromise, sign-in problems, or risky data use. Defender checks for compromise, Intra for identity risk, and Purview for insider risk. IT teams can view these risks in the Microsoft 365 admin center.  
  • Security policy teams in Intra help IT and security set and enforce organization-wide policies for new agents in the admin center.  

* These features are in public preview (available for testing for all users but not final) and will remain so on May 1.  

Secure And Manage Agent Access 

Managed agents can pose serious risks, including unauthorized access to resources, excessive privileges, or misuse by malicious actors. With Microsoft Intra features in Agent 365, you can secure agent identities and control their access to resources.  

  • Agent ID gives each agent a unique Microsoft Intra identity tailored to its needs. This enables organizations to set trusted, scalable access policies, close unmanaged identity gaps, and align agent access with existing controls.  
  • Identity protection and conditions for agents expand current user policies to agents. These policies make real-time access decisions based on risk, device compliance with Microsoft Intune (a device management tool), and custom security settings for agents working for a user. They help prevent compromise and make sure agents cannot be misused by bad actors.  
  • Identity governance for agents lets identity leaders limit agent access to only the sources they need. Access packages can be set to match a subset of user permissions, and leaders can audit which access has been granted to agents.  

Prevent Data Oversharing And Ensure Agent Compliance 

Agent 365, powered by Microsoft Purview, provides strong data security and compliance for agents. It helps prevent agents from retrieving sensitive data, stops insider data leaks, and supports responsible data administration to meet global regulations.  

  • Data security and posture management provide admins with clear visibility into data risks for agents, enabling them to resolve issues before they become trouble. Problems  
  • Information protection enforces MACMA 365 data sensitivity labels to prevent sensitive data leaks for agents.  
  • Insider risk management now covers agents blocking and flagging risky agent interactions with sensitive data for security admins.  
  • Data lifecycle management lets you set rules for keeping or deleting prompts and agent-generated data, helping you manage risk and liability.  
  • Audit and e-discovery now include agents enabling organizations to audit, investigate, and manage agent activity as they do for users and apps.  
  • Communication compliance now extends to two-agent interactions, enabling human monitoring of risky AI communications. This gives businesses and business leaders the ability to apply their code of conduct and compliance policy to AI as well, in advance of emerging cyber threats. Agent 365 includes Microsoft Defender protections purpose-built to detect and counter AI-specific vulnerabilities and threats such as prompt manipulation, model tampering, and agent-based attack chains.  
  • Security posture management for Microsoft Foundry and Copilot Studio agents identifies misconfigurations and vulnerabilities, so security twins can fix them before attackers exploit them.  
  • Detection in investigation and response for Foundry and Co-pilot agents helps teams investigate and fix attacks on agents, ensuring agents are included in security reviews, threat protection, and investigations and hunts using Agent 365. Tools Gateway helps organizations detect, block, and investigate malicious agent activities.  

Agent 365 will be available starting May 1, 2026, at $15 each per user per month. Learn more about Agent 365  

Microsoft 365 E7: the Frontier Suite 

Microsoft 365 E7 combines intelligence and trust to help organizations speed up frontier transformation. It provides human employees with AI tools for email, documents, meetings, spreadsheets, and business apps, and gives IT and security leaders the oversight and control needed for enterprise AI, including both users and AI agents.  

Microsoft 365 E7, Controllers Copilot, Agent365, Intra Suite, and E5 with Advanced Defender, Intra Intune, and Purview security features. It protects both users (humans) and AI agents. You can buy it starting May 1, 2026, for $99 per user each month. Learn more about Microsoft 365 E7.  

End-to-End Security for the Agentic Era 

Frontier transformation relies on agents and trust, which begins with security. Microsoft Security protects 1.6 million customers at AI speed and scale with Agent 365. These enterprise-grade tools now help organizations monitor, secure, and manage AI agents, offering full protection for both AI agents and human users with Microsoft 365 E7.  

Start your frontier transformation now with Agent 365 and Microsoft 365 E7: the frontier suite. Join us at the RSAC Conference 2026 to learn more about these solutions and hear from experts and customers who are molding the future of Asian Security.  

To learn more about agent security, visit our website, bookmark our security blog for expert updates, and follow us on LinkedIn (Microsoft Security) and X(@MSFTSecurity) for the latest cybersecurity news.  

Source: Secure agentic AI for your Frontier Transformation  

OpenAI acquires TBPN to expand AI Media Reach

We are excited to acquire TBPN, a team known for editorial talent, audience insight, and strong culture.  

TPPN is a place for real conversations about AI and its creators. Many of you already use the show to stay updated.  

Traditional communication doesn’t suit OpenAI. Leading tech change means fostering open conversations about AI’s impact, especially for builders and users.  

TBPN has already built this kind of space rather than trying to replicate it. It made sense to bring them in, support their work, and help them grow while keeping what makes them special. Going forward, TBPN will serve as the central hub for our dialogue with the AI community, fostering open conversation and sharing perspectives about the future of technology. Editorial independence is important. TBPN will continue to run its own programming, choose its guests, and make its own editorial decisions. This is essential to their credibility, and we are committed to protecting it in this agreement.  

We look forward to leveraging TVPN’s communication and marketing expertise to demonstrate how AI is changing daily life in new ways.  

TBPN will join our strategy group and report to Chris Lehane. Welcome, Jordi, John, Dylan, and team.  

A statement from TBPN:  

  • Over the past year, we’ve had a front-row seat not just to open air but to the entire ecosystem, covering daily news announcements and launches in real time. While we’ve been critical of the industry at times, after getting to know Sam and the OpenAI team, what stood out most was their responsiveness to feedback and their dedication to getting this right. Moving from commentary to real impact on how this technology is distributed and understood globally is incredibly important to us – Jordi Hays, co-founder and co-host of TBPN.  

About TBPN 

Technology, business, programming, network (TBPN) is a daily life tech show and one of the fastest-growing media companies. It is hosted by entrepreneurs Jordi Hayes and John Coogan on weekdays from 11 to 2 p.m. PT. The New York Times recently described TBPN as Silicon Valley’s newest obsession, led by Kogan Hayes and President Dylan Abruscato. TBPN caught fire across the technology ecosystem and can be found on X, YouTube, Spotify, Apple Podcasts, LinkedIn, Substack, and Instagram. 

Source: OpenAI acquires TBPN

Intellia Expands Hiring Push for AI-Driven Gene Editing 

Intellia Therapeutics is growing its team to accelerate the development of AI-powered and advanced gene-modification technologies. The goal is to strengthen its standing as a leader in CRISPR-based treatments. The company is also adding new DNA-writing tools and improving its delivery of gene editing in living organisms.  

Key Areas for Hiring and Growth 

  • Intellia is hiring to boost its AI and machine learning for new CRISPR-based therapies. This effort includes improving the accuracy of gene editing and delivery.  
  • Gene editing core team. The company is looking for experts to lead its genomics and RNA labs. Their work will focus on developing new gene‑editing components, such as sgRNAs and mRNAs, and expanding Intel Eason editing tools.  
  • DNA writing technology: after acquiring Rewrite Therapeutics, Intellia is increasing its work on CRISPR-guided DNA polymerization. This procedure allows precise genome editing without inducing double-strand breaks.  
  • In vivo and ex vivo development. New hires will help accelerate clinical trials for key programs focused on Transthyretin (ATTR) amyloidosis and hereditary angioedema.  

Strategic Overview 

  • Acquisition and pipeline growth: buying Rewrite Therapeutics in 2022 helped Intellia expand its technical abilities. The deal brought in new tools for precise DNA editing and helped address challenges in editing non-dividing cells.  
  • Regulatory milestones, even amid recent regulatory holds and trials in Talia, are advancing, with patient enrollment completed in its Phase 3 HALEO trial.  
  • Intellia is expanding its pipeline through partnerships with Regeneron and ONK Therapeutics. Together, they aim to create new gene-editing delivery technologies.  

Looking ahead, Telia’s growth is evolving beyond basic gene knockouts, with the company now targeting more complex genetic changes to treat a broader range of serious diseases.  

Intellia Therapeutics, a leading CRISPR gene-editing company, has stepped up hiring to strengthen its team of specialists. In early April 2026, the Cambridge-based biotech announced it was bringing in new experts in computational biology, machine learning, and auto-consciousness. Automated genomic analysis. This step follows the company’s clearance of several regulatory hurdles, including the end of a federal clinic hurdle on its main MAGNITUDE Phase 3 trial. By focusing on hiring people skilled in both data science and molecular biology, Intellia is moving toward an intelligence-driven approach to drug discovery. The goal is to move beyond manual experiments and use advanced predictive tools to identify the best genetic targets and improve how treatments are delivered for complex inherited diseases.  

A New Focus on Computational Biology 

Building on this focus, Intellia’s main goal with its current hiring is to add predictive genomic modeling to its research and development pipeline and investment, anticipated to yield robust returns by optimizing resource allocation. In the past, finding the safest and most effective places to edit genes, a process called off-target analysis, meant running countless lab tests on thousands of sites. Now the new computational experts will use high-quality simulations to replace much of this manual work, potentially accelerating R&D timelines and cutting costs. By bringing in people who can create digital pin models of human DNA Intel EXP experts to better predict CRISPR-T treatment outcomes across the whole genome before making any doses, minimizing risk, and potentially improving patient outcomes, both of which are key value drivers for stakeholders.  

Switching to a data-driven approach is fundamental for growing Intellia’s in vivo platform, which delivers gene-editing tools directly into the body. Treatments are so complex that they require more precision than humans can achieve on their own. The new computational scientists will work to improve the guide RNA sequences that direct the editing tools to the correct site. Using large datasets from earlier clinical trials. These hires will help Telia improve its delivery systems, especially its lipid nanoparticle (LNP) technology to ensure the treatment reaches the intended organ with high accuracy.  

Expanding Clinical Operations and Global Regulatory Teams 

Intellia is expanding its clinical operations and regulatory affairs teams to support more late-stage trials. As the Phase 3 MAGNITUDE trial approaches enrollment goals in early 2026, the company needs robust systems to manage data from multiple global sites. New hires in data management and biostatistics will process live participant data. This growth is essential as Intellia prepares to submit a biologics license application (BLA) for its hereditary angioedema (HAE) program later this year.  

The expansion in regulatory affairs goes hand in hand with growth in clinical operations. Genome medicine regulations are increasingly complex, so Intellia needs specialists who can track evolving global guidelines. The company is hiring policy architects to ensure its automated discovery meets FDA and EMA transparency standards. These experts document and validate computational target selection, linking rapid data processing with strict safety standards.  

Infrastructure And Extensive Genomic Manufacturing 

This focus on regulatory compliance directly informs investments in manufacturing capacity. As Intellia prepares for its first commercial launch in 2027, it is hiring more smart manufacturing staff, including process engineers and automation specialists, to develop modular production facilities. Gene editing therapies require a more flexible approach than traditional batch processing. New hires will implement automated quality control systems to monitor, in real time, the purity and strength of LNP-based therapies.  

Intellia uses detectors and automated feedback in production to ensure every vial meets strict clinical standards. This factory of the future can scale quickly if therapies gain broad regulatory approval. Automation boosts reliability and reduces manufacturing costs, making one-time treatments more accessible globally.  

Competitive Recruitment In The Biotech Corridor 

With expanded infrastructure, talent hiring is vital. Intellia competes for top talent in the Boston–Cambridge Biotech Corridor. To attract experts, it uses inducement brands and equity packages, common for public firms seeking leaders from tech. Operational expertise is now as valuable as patients in biopharma. A Digital Molecular Science Center in Telia appeals to scientists wanting to use advanced computing for health.  

The New Architecture of Healing 

As new experts join the labs and offices of the genomic era, the medical field is quietly changing. The practice of healing is increasingly relying on data and logic, predicting and treating diseases before they appear. Soon, doctors and data scientists may work as one using predictive insights to support patient care. Over time, rare diseases may become less frightening as systems are ready to find cures for genetic errors. In the future, our health could be managed by biotechnology that protects us and respects both our intentions and our biology.  

Source: intelliatx News Release 

Planet Limits Satellite Data Access Under US Request 

Planet Labs will stop delivering high-resolution satellite images of Iran and other conflict areas at the request of the US government. On March 9, 2026, the company will share these images only in special cases, such as urgent missions or clear public interest.  

Here are the main points of the new restrictions:  

  • The area of interest to which the restrictions apply is Iran and other parts of the region affected by the current conflict.  
  • Duration: This policy will likely stay in effect until the conflict ends.  
  • Industry impact: other companies, like Vantor (formerly Maxar), are also implementing stricter controls on satellite imagery in areas where US and allied forces are active.  
  • Purpose: The goal is to protect US and allied operations from possible retaliation.  

This decision shows that governments are taking greater control over commercial open-source intelligence during conflicts.  

Planet Labs (NYSE:PL), a major provider of global satellite images, said on Saturday. It will stop sharing visuals of Iran and the wider Middle Eastern conflict area for an indefinite period.  

The California company explained that the move comes after the U.S. government asked all commercial providers to hold back data from the region to protect operational security.  

Strategic Information Blackout 

This new policy goes further than the 14-day delay put in place last month. Planet Labs will now hold back all images of the conflict area from March 9 onward, and the restriction will likely remain until the fighting ends.  

The US government wants to stop adversarial groups from using commercial data for activities such as target identification, weapon guidance, or missile tracking. These abilities have become easier for non-state groups and foreign militaries to access through private companies.  

Planet Labs told its customers it will move to a managed distribution model. The Pentagon did not comment on intelligence issues for now. Will be shared only on a case-by-case basis, either for mission-critical needs or when there is a proven public interest.  

This change shows the rising conflict between the comm+ 

ercial space industries, open-source values, and the practical needs of a regional war that has spread through Saudi Arabia, Kuwait, and Bahrain since February 28.  

Industry Impact and Regulatory Precedent  

The order seems to be affecting companies in different ways. Vantor (formerly Maxar Technologies) said it has not been contacted by the government but is already using its own enhanced access controls.  

Vantor’s rules limit who can buy new or existing images in places where U.S. and allied forces are active. Black Sky Technology Inc (NYSE: BKSY) has not yet said if it has received similar instructions from the government.  

Investors pointed out that the shutter control is a major regulatory risk for the earth observation industry. Government contracts usually provide a steady income, but being forced to withhold data can disrupt commercial subscriptions and limit transparency for media and academic groups.  

As the conflict intensifies, imaging companies may find it harder to profit from their frequent satellite imagery, as national security rules can override existing business deals. 

Source: https://www.planet.com/newsroom/managed-access-notice 

Alphabet Commits Billions to Secure US AI Infrastructure

Alphabet’s growing investment in AI and cloud infrastructure highlights how rising demand is straining the systems behind enterprise computing. Major providers are spending more on computing power, but supply is still limited because AI workloads are growing faster than new data centers can be built.  

Alphabet’s recent earnings call made this challenge clear. The company expects to spend between $175 billion and $185 billion this year, nearly twice last year’s level. Most of this money will go toward servers, data centers, and networking equipment to support AI and cloud services.  

This trend goes beyond Alphabet. Other major cloud providers are also investing heavily in AI infrastructure to keep up with demand from businesses using generative AI for analytics and automation. For customers, the key point is what these investments reveal about ongoing infrastructure limits.  

Infrastructure Strain Reveals the Pace of AI Adoption 

We’ve been sharply constrained even as we’ve been ramping up our capacity, Alphabet CEO Sundar Pichai told analysts. Obviously, our capex spend this year is with an eye towards the future.  

This limitation is important because businesses are now using AI for more than just pilot projects. AI is being used in real production work, customer service, data analysis, software development, and planning. These tasks need steady computing power, quick response times, and stable performance. If infrastructure cannot keep up, projects take longer, and costs may rise.  

Alphabet’s cloud business shows how demand for AI is driving revenue growth. The company said its cloud unit grew 48% over the past year, reaching $17.7 billion last quarter, while analysts expected strong results. This growth implies that businesses are now using AI more widely, not just testing it.  

Cloud Growth Shows Shifting Enterprise Priorities 

This change also influences how businesses pick cloud providers, capacity, global reach, and how well AI tools work together are now as important as price. Companies using AI need to know that their infrastructure can handle sudden increases in use and support work in different regions. Supply limits show even the biggest providers are still working to meet demand.  

Pichai said he expects these limits to last through the year, underscoring that AI infrastructure is still catching up with what businesses need.  

Competition among large cloud providers adds another factor. Each one is building more data centers, developing custom hardware, and creating software to improve AI performance. This gives businesses more choices, but it also raises questions about how well different systems work together and about long-term vendor plans.  

Alphabet’s efforts are closely linked to its Gemini AI platform, which the company says is being widely used by business customers. Pichai told analysts that Gemini now has 8 million paid users across thousands of companies. AI tools are also being added to core products like search and advertising, which depend on large-scale computing power.  

We are seeing our AI investments and infrastructure drive revenue and expansion across the board, Pichai said.  

Planning for Capacity in an AI-Heavy Cloud Market 

For business planners, it’s important to watch how AI adoption and infrastructure growth are linked. Providers are investing to meet today’s needs and prepare for new workloads such as AI-powered search, automated document handling, and data-driven decisions. Decision‑making pools that require strong computing power  

Spending this much on infrastructure suggests that AI devices and services will continue to grow for years to come. Building data centers, buying hardware, and upgrading networks all take a long time. Businesses planning for the long term should expect ongoing changes in pricing, availability, and service options as providers try to match demand with supply.  

Investors had mixed reactions to Alphabet’s spending plans. Some viewed the increased spending as a risk to short-term profitability, while others saw opportunity. The company’s shares moved significantly after hours before settling as markets weighed higher spending against revenue growth for business customers. These market swings matter less than the main message: large cloud providers expect demand for AI computing to keep rising. A key question for enterprises is how to plan around that reality. Capacity constraints can affect deployment timing, regional availability, and service pricing. Organizations expanding AI workloads may need to build more flexibility into rollout schedules and vendor relationships.  

Ultimately, Alphabet’s big spending makes clear that AI infrastructure is now central to cloud providers, not just a third project. Businesses must base cloud strategies on anticipating where computing power will be needed most and how quickly providers can scale to meet accelerating demand.