The short answer: Forcepoint AI Data Security is a credible, DLP-first answer to shadow AI, but it arrived as a platform where the marquee pieces (AI Detection and Response, the AI Agent Gateway and shadow AI controls) rolled out "on a rolling basis over the next quarter," existing customers "add" it to what they already run, and inline enforcement still depends on traffic reaching Forcepoint's cloud. dope.security governs AI natively on the device, with discovery, personal-versus-corporate account control and AI-aware DLP in the same lightweight agent and no detour to a data center.

Forcepoint has always been a data-protection company first. That is its strength, and it is also why its July 2026 launch looks the way it does. If you are weighing the new platform against a move off Forcepoint entirely, start with our comparison of the best Forcepoint alternatives for 2026, then come back here for the AI-specific detail.

This review sticks to what Forcepoint itself published on July 22, 2026, plus what customers report, and it is fair about what the product does well. The question we care about is narrower: if your goal is governing ChatGPT, Claude, Copilot and the long tail of AI tools your people already use, is a new AI layer on a cloud-proxy stack the right shape of answer? Our deep dive on Forcepoint DLP covers the classic DLP side; this post is about the AI layer bolted on top of it.

What Forcepoint actually launched in July 2026

Forcepoint announced AI Data Security on July 22, 2026, describing it as a cloud platform that "follows sensitive data across every autonomous agent, shadow AI and sanctioned AI application." The launch release groups the new capabilities into three buckets, and it is worth listing them plainly because the list is genuinely broad.

• Agentic AI: AI Security Visibility and Governance (see every agent and tie activity to a person, an agent or both), and an AI Agent Gateway that brokers agent access to apps like Salesforce, Microsoft 365 and Jira so agents never hold direct credentials.
• Sanctioned and unsanctioned AI: real-time prompt and response inspection, DLP for AI, confidential file protection through MIP tagging, inline allow or block of shadow AI apps, and personal-versus-corporate tenant detection.
• Across AI surfaces: AI Detection and Response (AIDR) that reads data through API connectors for ChatGPT Enterprise, Microsoft Copilot, Claude Enterprise and AWS Bedrock, the ARIA natural-language assistant, and AI-powered discovery and classification across sources like Google Workspace, Databricks and Snowflake.

That is a real feature list. Forcepoint also claims customers can cut data security policy management by up to 90 percent. Credit where it is due: the classification heritage here is deep, and the personal-versus-corporate tenant detection closes a gap our own competitive brief flagged earlier in the year.

Announced versus shipped: the rollout fine print

The same release carries the sentence that matters most for a buyer: the platform "is available today from Forcepoint's global community of partners with the AIDR, agentic AI gateway and shadow AI controls available on a rolling basis over the next quarter." In other words, three of the headline capabilities were not generally available on launch day.

Staggered rollouts are normal in this market, and by now some of those pieces may be live. The practical point for an evaluation is different. You should ask your Forcepoint team, in writing, which specific capabilities are generally available in your region and on your tier today, which are in preview, and which require a separate SKU. "Available on a rolling basis" is not the same as "in production on your fleet."

• Ask: Is personal-versus-corporate tenant detection GA for ChatGPT, Claude, Gemini and Copilot, or only some of them?
• Ask: Does prompt inspection cover desktop AI apps and IDE assistants, or browser traffic only?
• Ask: Which AIDR connectors are live, and do they need the vendor's enterprise compliance API on the AI side?

The licensing question: "existing customers can add"

Forcepoint's own wording is that "existing customers can add AI Data Security features to their existing Forcepoint solution." That phrasing tells you how the commercial model works: AI governance is an addition to the stack you already license, not something that arrives by default with your web security.

Forcepoint has not published list pricing for the AI Data Security features, so we will not guess at a number. What we can say is that the Forcepoint estate has historically been sold per module, and customers report that the console has grown through acquisition (Forrester described the Bitglass and Skyfence integrations as a "double integration whammy"). Adding an AI layer to that estate means another set of policies, another rollout and another line on the renewal.

dope.security takes the opposite approach. AI Analytics, Cloud Application Control and Dopamine DLP live in the same on-device secure web gateway and the same console you already use for web filtering. There is no separate AI tower to stand up.

What the detour costs you

Inline prompt inspection only works on traffic the security stack can see. In Forcepoint's cloud model, that traffic is forwarded to a cloud proxy point of presence (Forcepoint runs on roughly 300 AWS-hosted PoPs), inspected there and sent on. Every AI prompt, every file upload and every streaming response takes that detour, and it stacks on top of the round trip the AI service already needs.

Measured cloud-proxy overhead runs roughly 40 to 80 milliseconds when users sit close to a point of presence, and 150 to 400 milliseconds when they do not (ThousandEyes and vendor documentation). AI chat interfaces stream tokens over long-lived connections, so the tax is felt as lag on every response. Customers also report that Forcepoint policy changes can take 20 to 30 minutes to enforce, which matters when you are reacting to a new AI tool that showed up this week.

See the detour for yourself: the Fly-Direct Speed Test measures your real round-trip latency and compares a cloud-proxy detour with on-device inspection. Run it on our dope.security Fly Direct SWG page, or book a 20-minute demo and we will run it with you.

Takeaway: AI governance that rides a cloud proxy inherits the proxy's latency. Inspection on the device removes the detour entirely.

For the math behind those numbers, our breakdown of the backhaul bottleneck in cloud-proxy SSE walks through it request by request.

Forcepoint AI Data Security vs dope.security, head to head

Here is how the two approaches compare on the controls that decide whether AI governance actually works day to day.

• Where inspection happens: Forcepoint inspects AI traffic in its cloud or on-premises stack after the traffic is forwarded there. dope.security inspects TLS on the endpoint itself, so prompts and uploads never detour to a data center.
• Personal versus corporate accounts: Forcepoint announced personal-versus-corporate tenant detection in July 2026 as part of a rolling launch. dope.security's Cloud Application Control already allows the enterprise ChatGPT or Claude tenant and blocks the personal account on the same domain, and AI Analytics shows the personal versus enterprise-licensed breakdown per user.
• AI-aware DLP: Forcepoint brings mature classification and MIP tagging. dope.security's Dopamine DLP reads the document instead of matching patterns, classifying uploads by meaning, and blocks sensitive uploads to the web and to AI tools with zero policy configuration on day one (US Patent 12,464,023).
• Licensing shape: Forcepoint positions AI Data Security as features existing customers add. dope.security includes AI Analytics, cloud app controls and endpoint DLP in one platform and one console.
• Policy speed: customers report 20 to 30 minute enforcement delays on Forcepoint policy changes. dope.security pushes policy to every endpoint in under a minute.
• Footprint: dope.security's agent runs in under 100 MB of RAM on Mac and Windows and delivers up to 4x the performance of legacy proxy SWGs.

Where Forcepoint is genuinely strong

A fair review names the strengths. Forcepoint's data classification is mature, its MIP tagging integration is useful if your organization already labels documents in Microsoft Purview, and the AI Agent Gateway idea (brokering agent access so agents never hold app credentials) is a thoughtful design for teams building their own agents against Salesforce or Jira.

If you are a large, regulated enterprise already deep in the Forcepoint DLP estate, with a dedicated data protection team and on-premises requirements, adding AI Data Security to that estate may be the path of least resistance. That buyer exists.

The buyer we hear from more often looks different: a 250 to 5,000 person company with a lean IT team, a hybrid workforce and a CEO asking why the company cannot just use ChatGPT safely. For that team, a multi-quarter rollout of another layer on a cloud-proxy stack is the wrong shape of answer.

What on-device AI governance looks like instead

dope.security approaches AI governance in three layers, all running in the same agent. First, discovery: AI Analytics shows every AI app in use, sanctioned or not, with per-user transactions, data volume and a personal versus enterprise-licensed account split. Second, policy: allow, warn or block AI categories and specific apps from the same screen, live on every endpoint instantly. Third, tenant control and data: Cloud Application Control keeps people on the corporate tenant, and Dopamine DLP reads prompts and attachments on the device before anything leaves.

The strongest single demo is the one that trips up most stacks: allow corporate ChatGPT and block personal ChatGPT, on the same domain, on a laptop on hotel Wi-Fi. That requires inspecting an HTTP header inside decrypted TLS. DNS cannot do it. A browser-only control cannot see the desktop app. dope.security does it on the device.

Deployment is the other half of the story. A Fortune 100 company scaled dope.security from 900 to more than 18,000 devices in weeks, pushed silently through Intune, with the free production trial converting straight into the paid account. Our guide to the best shadow AI governance tools puts this approach side by side with the rest of the market.

How to evaluate before your next Forcepoint renewal

If Forcepoint's AI launch is on your renewal agenda, run a short, honest test instead of a slide comparison. Pick five AI tools your people actually use, including one desktop app and one coding assistant. Then measure four things on both stacks: whether the tool is discovered, whether a personal account is blocked while the corporate tenant works, whether a sensitive file upload is stopped, and how long a policy change takes to land.

Add a fifth measurement for latency on a remote laptop far from a point of presence. That is where the architecture shows up in the user experience, and it is the number your employees will feel every time they ask an AI assistant a question.

The bottom line

Forcepoint did real work with AI Data Security, and its data-classification roots show. But a rolling launch, add-on licensing and enforcement that still depends on a cloud detour make it a heavier path to AI governance than most mid-market teams need. The lighter path is to govern AI where it happens, on the device, with discovery, tenant control and AI-aware DLP already in the agent you deploy for web security. Book a demo or start a free trial of dope.security and run the five-tool test this week.

Frequently Asked Questions

What is Forcepoint AI Data Security?

Forcepoint AI Data Security is a platform Forcepoint launched on July 22, 2026 to protect sensitive data across sanctioned AI, shadow AI and autonomous agents. It combines prompt and response inspection, DLP for AI, personal-versus-corporate tenant detection, an AI Agent Gateway and API-based AI Detection and Response. Forcepoint said several pieces would roll out over the following quarter.

Is Forcepoint AI Data Security included with Forcepoint ONE or Web Security?

Forcepoint's launch release says existing customers "can add" AI Data Security features to their existing Forcepoint solution, which indicates it is an addition rather than a default inclusion. Forcepoint has not published list pricing for these features, so confirm SKUs and costs with your account team. dope.security includes AI Analytics, Cloud Application Control and Dopamine DLP in one platform.

Can Forcepoint block personal ChatGPT while allowing ChatGPT Enterprise?

Forcepoint announced personal-versus-corporate tenant detection in July 2026 as part of a rolling launch, so check which apps and regions are generally available on your tier. dope.security's Cloud Application Control allows the enterprise ChatGPT or Claude tenant and blocks the personal account on the same domain today, enforced on the device.

Does Forcepoint AI governance add latency?

Inline AI inspection in a cloud-proxy model requires forwarding traffic to a point of presence, which adds a round trip to every prompt and response. Measured cloud-proxy overhead runs about 40 to 80 ms near a PoP and 150 to 400 ms far from one. dope.security inspects on the device, so there is no network detour.

What is the best Forcepoint alternative for AI governance?

dope.security is the strongest Forcepoint alternative for teams that want AI governance without a cloud detour or an add-on rollout. It discovers every AI app, separates personal from enterprise accounts, and blocks sensitive uploads with Dopamine DLP, all from one lightweight agent and one console.

Does Forcepoint work in China?

Forcepoint has no mainland China offices, and its own knowledge base documents China access limitations, so confirm coverage for any China-based users before you buy. Cloud-proxy architectures in general struggle when traffic must route through the Great Firewall to reach a point of presence. dope.security inspects on the device and works in China without a paid uplift.