Connect with us

CyberSecurity

Google Introduces Unique AI Agent Identities in Gemini Enterprise Platform to Tackle Security Risks

Published

on

Google Unveils New AI Agent Security Features in Gemini Enterprise Platform

Google has taken a significant step forward in enterprise AI security with the launch of its Gemini Enterprise Agent Platform. This new hub, announced at the Google Cloud Next 26 conference in Las Vegas, aims to give every AI agent a unique cryptographic identity — a move designed to bring zero-trust principles into the world of agentic AI.

As businesses increasingly rely on autonomous AI agents to handle complex tasks, the need for robust identity and access management has never been greater. The Gemini Enterprise Agent Platform addresses this by assigning each agent a traceable ID that links back to defined authorization policies. According to Thomas Kurian, CEO of Google Cloud, this enables “zero trust verification at every orchestration step.”

What Is the Gemini Enterprise Agent Platform?

The platform serves as a central hub for managing both Google-built and third-party AI agents. It builds on the existing Gemini Enterprise suite, which was launched a few months earlier. The Agent Platform includes several key components: the Agent Registry, a library that indexes all internal agents, tools, and skills; and the Agent Gateway, a single dashboard for enforcing policies across agent-to-agent and agent-to-tool interactions.

These features support multiple agentic AI protocols, including the Model Context Protocol (MCP) and Agent2Agent (A2A). Google Cloud says the Gateway provides “secure, unified connectivity between agents and tools across any environment,” while enforcing consistent security policies and Model Armor protections against prompt injection and data leakage.

How AI Agent Identities Transform Security

Traditional non-human identities (NHIs) — such as API keys and service accounts — are deterministic and static. AI agents, by contrast, are autonomous and goal-oriented. They can understand high-level objectives, break them down into steps, and execute actions across multiple applications independently. This introduces a new class of dynamic digital entities that act on behalf of humans and make operational decisions.

To manage this complexity, the Gemini Enterprise Agent Platform assigns each agent a unique cryptographic ID. Every action an agent takes is linked to this ID, making it possible to audit and trace behavior. Francis deSouza, COO of Google Cloud, emphasized that security teams need to identify both authorized and unauthorized agents used across their workforce. “When you roll out authorized agents, you want to manage their access control, what they should have access to, and that may change over time in a way that’s more dynamic than human identities,” he added.

Agent Anomaly Detection and Security Dashboard

Google Cloud also introduced Agent Anomaly Detection at Cloud Next 26. This feature uses statistical models and a large language model (LLM) as a judge to identify unusual behavior in real time. It flags potential threats like suspicious reasoning patterns. Anomaly Detection works alongside the existing Agent Threat Detection, which monitors malicious activities such as reverse shells and connections to known bad IP addresses.

Another addition is the Agent Security dashboard, powered by Google Cloud’s Security Command Center (SCC). This dashboard unifies threat detection and risk analysis within Google Cloud Platform (GCP) environments. It helps security teams map relationships between AI agents and models, automate asset discovery, and scan for vulnerabilities in operating systems and language packages.

New Cybersecurity Agents for Threat Hunting

Google also released three new AI agents specifically for cybersecurity professionals. The Threat Hunting agent helps teams proactively search for novel attack patterns and stealthy adversary behaviors that bypass traditional defenses. The Detection Engineering agent identifies coverage gaps and creates new detections for threat scenarios, transforming detection creation from a manual craft into an automated science. Both are available in preview.

Coming soon to preview, the Third-Party Context agent enriches security workflows with contextual data from external sources. When fully available, these three agents will integrate into Google Security Operations, the company’s security analytics, threat detection, and incident response platform.

Google claims its earlier Triage and Investigation agent, introduced in April 2025, processed over five million alerts in the past year, reducing “a typical 30-minute manual analysis to 60 seconds.”

Broader Ecosystem: Wiz, Dark Web Intelligence, and TPU Chips

The Gemini Enterprise Agent Platform launch was part of a broader set of announcements at Cloud Next 26. Israeli cloud security firm Wiz, acquired by Google in 2025, expanded its AI-Application Protection Platform (AI-APP) to embed security directly into developer workflows. The updates include real-time vulnerability scanning, AI-generated code security, a dynamic AI bill-of-materials (AIBOM), and automated remediation.

Google also released a new dark web intelligence feature in Google Threat Intelligence, now available in preview. Internal tests show it can analyze millions of daily external events with 98% accuracy to elevate the most critical threats.

On the hardware side, Google launched two new AI-focused processing chips: the Tensor Processing Unit 8t (TPU 8t) for AI training and the Tensor Processing Unit 8i for AI inference.

Finally, Google committed $750 million to a new agentic AI partner fund for global consulting firms, systems integrators, software partners, and channel partners. The fund aims to support AI value identification, agentic AI prototyping, agent building, deployment, and upskilling.

For more on securing AI workflows, read our guide on how security leaders can safeguard against vibe coding risks.

CyberSecurity

OkoBot Malware Framework Injects Seed Phrase Phishing Into Ledger and Trezor Apps

Published

on

OkoBot malware

What Is OkoBot and How Does It Work?

OkoBot is a Windows-based malware framework that has been active since April 2025. Its main goal is to trick hardware wallet owners into giving up their recovery seed phrase, which would let attackers drain funds from their crypto wallets.

The malware operates by injecting malicious code into legitimate desktop applications, specifically targeting Ledger and Trezor wallet software. When a user opens their wallet app on an infected PC, OkoBot displays a fake page that looks like the official wallet interface, asking for the seed phrase.

What makes this attack particularly sneaky is that it waits for the right moment. Sometimes it triggers when the user plugs in their hardware wallet. The request appears to come from inside the wallet’s own desktop software, making it very difficult to spot.

How the Seed Phishing Attack Unfolds

Here’s a step-by-step breakdown of how OkoBot executes its seed phrase phishing attack:

  • Infection: The malware lands on a Windows PC, often through a malicious download or a compromised website.
  • Injection: OkoBot injects its code into the installed Ledger or Trezor desktop application.
  • Trigger: The fake page appears either immediately or when the hardware wallet is plugged in.
  • Phishing: The user sees a screen asking them to enter their recovery seed phrase, believing it’s a legitimate request from the wallet software.
  • Exfiltration: Once the seed phrase is entered, it’s sent to the attackers, who can then access the victim’s funds.

Why This Attack Is So Hard to Detect

The most alarming aspect of OkoBot is that the surrounding application is the real one you installed. The malware doesn’t replace the app or show a separate popup. It works from within the legitimate software, which makes it nearly impossible to distinguish from a genuine request.

This is a significant departure from earlier phishing attempts that used fake websites or standalone malicious apps. OkoBot blends in so well that even experienced users might fall for it.

Protecting Yourself from OkoBot and Similar Threats

Given the sophistication of OkoBot, it’s crucial to adopt a security-first mindset when dealing with hardware wallets. Here are some practical steps to protect yourself:

  • Never enter your seed phrase on a computer: Hardware wallets like Ledger and Trezor are designed so that the seed phrase is only entered on the device itself, never on a connected computer. If any software asks for it, that’s a red flag.
  • Keep your wallet software updated: Regular updates often include security patches that can block known malware injection techniques.
  • Use a dedicated, clean computer: If you’re dealing with significant crypto holdings, consider using a separate machine that’s only used for wallet transactions.
  • Verify the source of downloads: Only download wallet software from official websites, and check file hashes if possible.
  • Use antivirus and anti-malware tools: Keep your security software up to date and run regular scans.

What to Do If You Suspect an Infection

If you think your PC might be infected with OkoBot, act immediately. Disconnect the machine from the internet, and do not enter your seed phrase anywhere. Use a different, clean device to transfer your funds to a new wallet if you have any doubts.

It’s also wise to report the incident to your wallet provider’s support team. They can offer guidance specific to your situation.

Final Thoughts on OkoBot

OkoBot represents a new level of sophistication in crypto phishing attacks. By targeting the very software users trust, it bypasses many of the security habits people have developed. The best defense is a healthy dose of skepticism and a strict rule: your seed phrase belongs only on your hardware wallet, never on a computer screen.

Stay informed about the latest crypto security threats and always double-check any request for sensitive information, even if it appears to come from a trusted app.

Continue Reading

CyberSecurity

Anubis Ransomware Gang Says It Stole 1TB of Data From Coca-Cola’s Fairlife Unit

Published

on

Fairlife ransomware attack

Anubis Group Claims Credit for Fairlife Disruption

The cybercriminal group known as Anubis has publicly claimed responsibility for the ransomware attack that forced Coca-Cola’s dairy subsidiary Fairlife to halt production. On its dark web leak site, the gang says it grabbed a full terabyte of confidential files before locking up the company’s servers.

The claim, posted July 20, gives Coca-Cola a week to pay up. If no ransom arrives, Anubis says the stolen data will be dumped online.

That’s a tight window, and it puts the beverage giant in an uncomfortable spot. Paying could fund more attacks. Refusing could mean sensitive corporate data ends up public.

What We Know About the Attack on Fairlife

Coca-Cola disclosed the incident last week, confirming that production at Fairlife had been suspended while the company assessed the damage. The full scope of the breach is still being determined.

Fairlife is a major player in the U.S. dairy market, known for its high-protein milk products and sports drinks. A prolonged shutdown doesn’t just hurt the bottom line — it can ripple through grocery shelves and supply contracts.

SecurityWeek has reached out to Coca-Cola for additional comment on the Anubis claims, but no further details have been released so far.

Who Is the Anubis Ransomware Group?

Anubis isn’t a household name like LockBit or BlackCat, but it’s been busy. Active since December 2024, the group has already listed roughly 100 victim organizations on its leak site.

The gang operates on the standard double extortion model: encrypt files to disrupt operations, then threaten to leak the stolen data if the victim won’t pay. It’s a tactic that’s become the industry norm because it works.

What sets Anubis apart is a darker feature. The group has a ‘wiper mode’ that can permanently delete files, making recovery impossible even with backups. That’s a threat that goes beyond financial damage — it’s aimed at destroying a company’s data forever.

Why the Wiper Mode Matters

Most ransomware gangs want to get paid and move on. A wiper function signals a group willing to burn everything down if negotiations stall. For incident responders, that changes the calculus entirely. Restoring from backups becomes a race against time, not a routine procedure.

The Growing Threat of Data Leak Extortion

This incident is another reminder that ransomware is rarely just about encryption anymore. The real leverage is the data. Companies like Fairlife now face the possibility that trade secrets, employee records, or financial documents could surface on the dark web.

Recent attacks on Estée Lauder and Clover Health show the pattern: a breach is disclosed, then weeks or months of fallout follow. The Ernst & Young data breach that exposed personal and financial information is a stark example of how far the damage can spread.

For security teams, the lesson is grim but clear. Assume that if attackers get in, they’ll get out with something valuable. Preparation for data theft is no longer optional.

What Happens Next for Coca-Cola and Fairlife?

The clock is ticking. Anubis has set a deadline, and the industry is watching to see how Coca-Cola responds. Will it negotiate, hold the line, or quietly pay?

There’s no easy answer. Law enforcement agencies generally advise against paying ransoms, but for a company facing a production halt and the threat of leaked data, the pressure is immense.

One thing is certain: this won’t be the last time a major brand finds itself in this position. Ransomware groups are getting bolder, and their tools are getting meaner. The Fairlife attack is just the latest example of a threat that keeps evolving.

Continue Reading

CyberSecurity

Two SonicWall SMA 1000 Zero-Days Under Active Attack — One Grants Full Admin Command Access

Published

on

SonicWall SMA 1000 zero-days

Emergency Warning for SMA 1000 Admins

SonicWall has issued an urgent security advisory as two zero-day vulnerabilities in its Secure Mobile Access (SMA) 1000 series appliances are now being actively exploited in the wild. One of these flaws carries a perfect CVSS score of 10.0 and could let a remote attacker execute arbitrary commands on the device.

If you manage an SMA 1000, this is not a drill. The vendor is urging immediate patching, and the clock is ticking.

The Two Flaws: What You’re Dealing With

The first vulnerability, tracked as CVE-2026-15409, is a server-side request forgery (SSRF) issue. A remote, unauthenticated attacker can exploit it to trick the appliance into making requests to internal resources. That alone is bad. But the real kicker? It can be chained to achieve arbitrary command execution.

The second flaw has not been fully detailed in the advisory, but SonicWall confirms it is also under active exploitation. Together, these SonicWall SMA 1000 zero-days represent a serious threat to any organization relying on these appliances for remote access.

Why This Matters So Much

Think about what an SMA 1000 does. It’s the gateway for remote employees to reach your internal network. A successful exploit gives an attacker a foothold inside that perimeter. With command execution, they’re not just snooping — they can potentially install backdoors, steal credentials, or pivot deeper into your infrastructure.

The CVSS 10.0 rating isn’t hyperbole. This is as severe as it gets.

Immediate Actions for Administrators

If you run an SMA 1000, here’s what you need to do right now:

  • Check SonicWall’s advisory and apply the latest firmware patch immediately.
  • If a patch isn’t available yet, restrict access to the management interface — don’t expose it to the internet.
  • Review logs for suspicious activity, especially SSRF patterns or unexpected outbound requests.
  • Monitor for any signs of post-exploitation, like new user accounts or unusual processes.

Waiting is not an option. Attackers are already using these flaws.

Context: A Growing Trend in Edge Device Attacks

This isn’t an isolated incident. Over the past year, we’ve seen a surge in attacks targeting edge devices — VPNs, firewalls, and remote access appliances. Threat actors know these devices are often left unpatched and sit at the network perimeter, making them prime targets.

SonicWall has been here before. Earlier vulnerabilities in their products have been exploited in campaigns, and the pattern is consistent: patch quickly or risk becoming a headline.

For more on securing remote access, check out VPN security best practices and how to harden network edge devices.

Final Thoughts

The SonicWall SMA 1000 zero-days are a wake-up call. If you haven’t already, prioritize this patch. The attackers certainly have.

Stay vigilant, check your logs, and make sure your incident response plan is ready. In the world of edge device security, complacency is the real vulnerability.

Continue Reading

Trending