Vendor, ranked 5 of 15

HUMAN Security

In short

HUMAN Security sells bot and agent controls that grade yes on agent verification, per-purpose policy and reporting and partial on charging, for 7 of 10 points. Cryptographic verification is missing at the medium trust level, which HUMAN itself flags as a source of false negatives. The price is not public.

Bot and fraud defense vendor whose AgenticTrust module rates AI agents High, Medium or Low trust and lets sites allow or deny agent actions such as browsing, login, account creation and checkout. Vendor description of itself, checked against its documents on 2026-11-04.

Alternatives to HUMAN Security · humansecurity.com

The rubric, with evidence

Cost

no public price; sales-led (checked 2026-11-04) Source, read 2026-11-04.

What HUMAN Security does not do

• Medium-trust agents lack cryptographic verification, so HUMAN itself warns of false negatives (Trust Levels blog, 2026-09-02).
• No native pay-per-crawl; monetization is through partner tools (AgenticTrust page).
• No published price list (package plans page).
• HUMAN does not charge AI crawlers itself; the AgenticTrust page points to scraping monetization partners.
• HUMAN's own docs say agents without cryptographic verification can be misread as human users, so they stay at Medium trust.

Documented facts

Features are as the vendor documents them; we have not tested them. A claim a vendor makes about itself is marked as a vendor claim.

• Primary source AgenticTrust is a module of HUMAN Sightline Cyberfraud Defense; Confused.com integrated it for its insurance journeys. Source, read 2026-11-04.
• Primary source HUMAN assigns observed agents a High, Medium or Low Trust Level; all agents start at Medium, and High goes to agents verifiable by a cryptographic key on each request. Source, read 2026-11-04, dated 2026-09-02.
• Primary source Trust Levels weigh evidence of abuse or bypassing controls, cryptographic identity, behavioral consistency and discoverability, and a High level can be lowered if an agent starts misbehaving. Source, read 2026-11-04, dated 2026-09-02.
• Primary source Customers set agent permissions per category such as Content and Products, Engage, Account Creation, Checkout and Rate Limit, keyed to trust level. Source, read 2026-11-04, dated 2026-09-02.
• Vendor claim The AgenticTrust page says sites can deny or allow browsing, login or checkout per agent and verify agents with cryptographic signatures. Source, read 2026-11-04.
• Vendor claim HUMAN's Satori team found 5.7% of traffic presenting an AI crawler or scraper user agent was spoofed. Source, read 2026-11-04, dated 2026-09-02.
• Vendor claim HUMAN reports 7,851% year-over-year growth in agentic AI traffic since 2025. Source, read 2026-11-04, dated 2026-09-02.
• Vendor claim In August 2026 agentic traffic grew 27% month over month with a 21.9% blocking rate; Comet Browser was 40.7% of agentic traffic, Claude Chrome Extension 22% and ChatGPT Agent 19.8%. Source, read 2026-11-04, dated 2026-09-03.
• Primary source HUMAN offers cryptographic verification of the Amazon Bedrock AgentCore Browser so agents built on AWS can be told apart from spoofed ones. Source, read 2026-11-04.
• Vendor claim The AgenticTrust page lists integrations with AI developers, chargeback protection vendors, scraping monetization tools, identity providers, OWASP and AGNTCY.org. Source, read 2026-11-04.
• Vendor claim HUMAN says it was named a Leader in The Forrester Wave: Bot and Agent Trust Management Software, Q2 2026. Source, read 2026-11-04, dated 2026-06-15.
• Primary source HUMAN's package plans page lists Sightline AgenticTrust without any published price. Source, read 2026-11-04.

Dated news

• 2026-09-02HUMAN publishes Agent Trust Levels for AgenticTrust. HUMAN described its High, Medium and Low Trust Levels for AI agents. Agents start at Medium, move to High with per-request cryptographic verification and good behavior, and drop to Low for spoofable identity or abuse. Levels drive default permissions for checkout, account creation and similar actions. Source.
• 2026-09-03HUMAN State of Agentic Traffic, August 2026. HUMAN's monthly benchmark reported agentic traffic up 27% month over month with 21.9% blocked. Comet Browser led at 40.7% of agentic traffic, Claude Chrome Extension held 22%, and ChatGPT Agent rose to 19.8%, with Codex appearing for the first time. Source.

Compared with

• DataDome vs HUMAN Security: DataDome publishes plan prices (from USD 3,830 a month); HUMAN publishes none and rates agents High, Medium or Low trust, with High reserved for agents that carry a cryptographic key on each request.
• HUMAN Security vs Kasada: Kasada launched AI Agent Trust on 2026-01-21 with a verified agent directory that supports Web Bot Auth; HUMAN's AgenticTrust grades agents by trust level and per-action permissions, without naming Web Bot Auth on the pages read.
• Akamai vs HUMAN Security: Akamai is listed by OpenAI as recognising ChatGPT's Cloud browser under the bot name "ChatGPT Agent"; HUMAN lists the same agent in Known Bots & Crawlers and AgenticTrust permissions.
• Cloudflare vs HUMAN Security: Cloudflare identifies ChatGPT's Cloud browser by bot tag chatgpt-agent and detection ID 129220581; HUMAN identifies it as "ChatGPT Agent" and adds AgenticTrust permissions per action.

Last reviewed 2026-11-05