Quick wins for a faster PC:
Clear out junk files and repair common Windows errorsFree Scan →Scan for outdated or missing drivers - takes under a minuteDriver Scan →F5 did not merely announce a planned $180 million purchase of CalypsoAI—it completed the transaction on September 26, 2025. The acquired technology now operates as F5 AI Guardrails and F5 AI Red Team, which became generally available on January 14, 2026. F5 announced $180 million in purchase consideration, while its subsequent filing reported $145.2 million in cash paid; those are different transaction measures, not interchangeable prices.
The deal, from announcement to products
| Date | What happened |
|---|---|
| September 11, 2025 | F5 announced its intent to acquire CalypsoAI for $180 million in purchase consideration, primarily financed with cash. F5 announcement |
| September 26, 2025 | The acquisition closed, according to F5’s fiscal 2026 first-quarter Form 10-Q. F5 Form 10-Q |
| September 29, 2025 | F5 introduced the acquired capabilities as F5 AI Guardrails and F5 AI Red Team. F5 product introduction |
| January 14, 2026 | F5 announced general availability for both products. F5 general-availability announcement |
That timeline matters because older coverage may still describe CalypsoAI as a pending target. It is now an F5 business and its technology is being sold under F5 product names.
What CalypsoAI brought to F5
CalypsoAI, founded in 2018 and operating primarily from Dublin, Ireland, focused on security at the AI inference layer: the interaction among users, applications, agents, models, tools and data. F5 said CalypsoAI had raised more than $40 million from investors including Paladin Capital Group, Lockheed Martin Ventures and Hakluyt Capital.
Its capabilities were positioned around:
- Real-time defense against prompt injection and jailbreaks
- Data-leakage prevention and runtime policy enforcement
- AI red teaming and adversarial testing
- Auditability and governance
- Model-agnostic protection across cloud and on-premises environments
F5 already sold application delivery, API security and traffic-management controls. Those controls do not automatically understand an AI prompt, a model output, an agent’s tool call or sensitive information embedded in a conversation. The acquisition was intended to fill that architectural gap, not to replace F5’s existing application-security products.
Recommended Free Tools
#1 Best Overall
- Compact and Efficient Design: The FortiGate 40F is designed for small to mid-sized businesses and enterprise branch offices, featuring a compact, fanless desktop form factor that ensures quiet operation and minimizes space usage.
- Robust Connectivity Options: Equipped with 5 GE RJ45 ports, including 1 WAN port and 4 internal ports, this model provides essential connectivity and flexibility for various network configurations in a small-scale environment.
- High-Performance Security: Offers up to 1 Gbps IPS throughput and 600 Mbps threat protection throughput, using Fortinet’s purpose-built security processor technology to deliver industry-leading performance and protection for SSL encrypted traffic.
- Advanced Threat Protection: Integrated with Fortinet’s AI-powered FortiGuard Labs, the FortiGate 40F offers comprehensive cybersecurity, identifying and mitigating both known and unknown threats to maintain robust security across your network.
- Simplified Management and Deployment: Features a user-friendly management console that provides comprehensive network automation and visibility, coupled with Zero Touch Integration with Fortinet’s Security Fabric for easy deployment.
What F5 AI Guardrails does
F5 describes AI Guardrails as runtime protection and governance for models, AI applications, agents, prompts, data and outputs. Vendor-stated capabilities include:
- Prompt-injection and jailbreak detection and blocking
- Data-exfiltration and sensitive-data-leakage controls
- Content moderation
- Custom policies written in natural language
- Controls for PII, PHI, PCI and EU AI Act-oriented requirements
- Agent and tool-call visibility
- Audit-ready logging
- Model-agnostic enforcement
- Deployment in public cloud, private cloud, on-premises and air-gapped environments
The product page also describes support for agents using OpenAI, Anthropic or similarly formatted interfaces. These are F5’s stated capabilities, not independent validation that every control performs identically across models, languages, frameworks or deployment patterns. Details are available on the F5 AI Guardrails product page.
What F5 AI Red Team does
AI Red Team is the testing component. F5 says it uses automated attacker agents to simulate prompt injection, jailbreaks and other attacks against models, applications and agentic workflows. It provides risk-scored findings, attack-path tracking and agent fingerprints, and F5 says its threat database adds more than 10,000 attack patterns monthly.
One important qualification appears in F5’s own FAQ: the product uses agents to run attack campaigns, but its targets are currently deployed models or applications rather than autonomous agents as a separate testing target. Buyers should therefore confirm how their particular agent framework, tool protocol and multi-step workflow are represented.
Rank #2
- HARDWARE PLUS SECURITY SERVICES: FortiGate-60F Firewall Appliance bundled with 1 year of FortiCare Premium and FortiGuard Unified Threat Protection.
- UNIFIED THREAT PROTECTION (UTP): Secures against advanced online threats with comprehensive web filtering and anti-botnet technologies.
- OPTIMIZED FOR MEDIUM-SIZED BUSINESSES: Tailored for businesses needing robust security without the infrastructure of larger enterprises.
- RELIABLE CUSTOMER SUPPORT: FortiCare Premium ensures high-quality support and service continuity.
- EFFECTIVE PROTECTION: Employs advanced filtering technologies to safeguard against sophisticated threats.
AI Red Team is described at F5’s product page.
How testing and runtime enforcement fit together
- Discover: AI Red Team launches adversarial campaigns against a deployed model, application or workflow.
- Prioritize: Findings are risk-scored and accompanied by attack paths or explanations.
- Design a response: Security teams create or activate policies addressing the finding.
- Enforce: AI Guardrails applies those policies to live prompts, outputs, data and tool calls.
- Review: Logs and subsequent tests support governance, tuning and regression checking.
F5 says findings can be translated into active guardrails, but that is not the same as a vulnerability being fixed automatically. Teams still need approval gates, false-positive testing, staged rollout, exception handling and rollback. F5’s integration concept is outlined in its AI-security platform material.
What “model-agnostic” means—and does not mean
F5 uses “model-agnostic” to mean that policies can be applied across public and private models rather than being tied to one provider. That is useful for organizations using several large-language-model vendors, open-source or fine-tuned models, private deployments, or frequent model changes.
It does not establish identical functionality for every model, agent protocol, streaming interface or tool-call pattern. During an evaluation, ask which interfaces are supported, how policies move between environments, what telemetry is available, and how much latency inspection adds.
The $180 million headline versus the $145.2 million filing
| Figure | What it represents | Source and qualification |
|---|---|---|
| $180 million | Purchase consideration announced by F5 | Reported in the September 11, 2025 announcement; primarily financed with cash. F5 announcement |
| $145.2 million | Cash reported after closing | Reported in F5’s fiscal 2026 first-quarter Form 10-Q for the September 26 closing. F5 Form 10-Q |
The available filing excerpt does not provide a full reconciliation between purchase consideration and cash paid. It is therefore inaccurate to call $145.2 million the announced purchase price, to call the entire $180 million a cash payment, or to describe the difference as a discount without transaction documentation supporting that conclusion.
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Scan for outdated or missing drivers - takes under a minute3Clear out junk files and repair common Windows errorsRank #3
- 【Up to 1100 Mbps VPN Speed 】 Hardware-accelerated WireGuard and OpenVPN-DCO deliver up to 1100 Mbps VPN throughput, over 3× faster than Brume 2 for smooth remote access and file transfers.
- 【Three 2.5G Ports & Multi-WAN】Tri-port 2.5GbE design with flexible WAN LAN configuration supports multi-gigabit wired setups, dual-ISP Multi-WAN and failover to keep home and SOHO networks online.
- 【Stealth VPN Obfuscation】VPN obfuscation disguises VPN traffic as regular HTTPS, helping you evade blocking, bypass restrictive networks and maintain stable, private connections.
- 【DPI protection】Deep Packet Inspection with visual dashboards blocks adult/gambling/malicious sites, while SQM and QoS prioritize gaming, calls, and video when bandwidth is tight
- 【OpenWrt & USB 3.0 Expansion】OpenWrt with 1GB DDR4 and 8GB eMMC lets you install plugins and build VPN, ad-blocking or NAS, while USB 3.0 Type‑C connects high-speed storage or 4G/5G dongles
F5 said the acquisition was expected to be immaterial to revenue and operating results, and its later filing likewise described CalypsoAI’s revenue and earnings as not material in the reported period. The near-term thesis was capability expansion and platform strategy, not an immediate revenue step-up.
Where the acquisition fits in an enterprise security architecture
AI Guardrails and AI Red Team address selected risks around inference. They do not replace identity and access management, secrets management, data classification, secure software development, API and application security, infrastructure monitoring, model and dataset governance, incident response or human approval for high-impact decisions.
Guardrails can inspect and block specified interactions; they cannot guarantee that an AI system is factually correct, unbiased, legally compliant or safe in every business context. Organizations should also test difficult operating conditions:
- Streaming responses where unsafe content may appear before full-response evaluation
- Multilingual, encoded or obfuscated attacks
- Indirect injection through documents, websites or retrieval results
- Agents accessing databases, browsers, SaaS tools or code interpreters
- Multiple models chained in one workflow
- Shadow AI applications bypassing an approved gateway
- Air-gapped environments with restricted threat-intelligence updates
- False positives in healthcare, finance, legal or customer-service workflows
- Regional policies that differ by country or business unit
Should an F5 customer consider it?
The strongest potential fit is an existing F5 customer seeking one relationship for application delivery, API security, AI runtime controls and testing—especially where private-cloud, on-premises or air-gapped deployment matters. F5’s positioning is described in its AI-security overview.
Rank #4
- Runs UniFi Network for full-stack network management
- Manages 30+ UniFi Network devices and 300+ clients
- 1 Gbps routing with IDS/IPS
- Multi-WAN load balancing
- 0.96" LCM status display
A specialist AI-security product may be preferable when a team needs narrower expertise, broader integrations or no dependence on the F5 platform. Cloud-provider-native controls can be simpler for a workload concentrated in one cloud, while open-source components offer customization at the cost of maintenance, threat-intelligence updates, support and auditability. A layered architecture can provide best-of-breed tools but creates more integration and policy-management work.
Customer due-diligence checklist
- Commercial scope: Is AI Guardrails included in an existing F5 Application Delivery and Security Platform entitlement? Is pricing based on models, requests, tokens, users, workloads or deployment?
- Performance: What latency and throughput impact does inspection create? Does the system fail open or fail closed during an F5 or model-provider outage?
- Data handling: Where are prompts, responses and logs stored? Can sensitive telemetry remain entirely in the customer environment?
- Deployment: Are private-cloud, on-premises and air-gapped features available in the required edition, with equivalent threat-intelligence updates?
- Integration: Which model providers, agent frameworks, streaming modes and tool-call protocols are supported?
- Operations: Can policies be version-controlled, exported and rolled back? How are exceptions, approvals and false positives handled?
- Testing: Can teams create custom and multi-turn attacks, reproduce findings and require human review before a test finding becomes a production policy?
- Support: What service-level agreements, implementation services and support tiers apply?
F5’s current product pages direct buyers to sales or demonstrations rather than publishing a standard list price for either product. Do not assume a trial or subscription call to action establishes an enterprise price or included functionality.
Bottom line
F5’s CalypsoAI acquisition added a credible AI-inference security layer to its broader application and API portfolio. The transaction closed, produced AI Guardrails and AI Red Team, and was financially immaterial in the near term by F5’s own description. Whether it is the right purchase depends less on the $180 million headline than on verified efficacy, supported interfaces, latency, privacy, pricing, operational controls and the customer’s existing F5 footprint.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
Free tools Windows power users keep installed
One-click scans. No signup required.
