Bot and fake-account defence for sign-up, login and checkout that reads 225+ risk signals and now sorts AI agent traffic into good and malicious.
Pick Arkose Labs if bots and fake accounts at sign-up and login are your biggest fraud cost and you run a large consumer platform. Its Bot Manager reads 225+ signals per session. It does nothing for payment fraud or chargebacks, and small apps will find it too heavy.
Challenges that change as attackers adapt, shown only to suspect traffic, plus allow, monitor or block policy for AI agents.
Its bot, device and AI agent products share one Arkose Titan session layer, per Arkose, so they are hard to unbundle.
Main trade-off
Customisation and list changes often run through Arkose's own team rather than self-service.
Arkose Labs does not publish prices. Arkose Titan, Bot Manager, Device ID and Agent Trust Manager are sold through a demo and an enterprise contract, and one user writing on G2 (seen 22 November 2026) calls it not a cheap solution.
Arkose Labs is a bot and account-abuse defence platform that protects sign-up, login and checkout by scoring each session and sending only suspect traffic adaptive challenges. Its Agent Trust Manager also classifies AI agents by intent, and Arkose says 97.7% of legitimate agent sessions it saw carried no verified disclosure.
Bot Manager
Judges each session from 225+ risk signals and the Arkose Global Intelligence Network, passing real users without friction.
Dynamic challenges
Suspect traffic meets challenges that change in real time, making bot farms and paid solvers more expensive to run.
Agent Trust Manager
Sorts AI agent sessions into three groups by intent and applies allow, monitor or block per agent and per step.
Device ID
Recognises returning devices across sessions, used in Arkose's SMS toll fraud case study.
Email intelligence and phishing protection
Scores sign-up emails and detects phishing kits that copy a brand's login page.
Arkose Command Center
One console for policy, dashboards and threat insight across every protected endpoint.
| Tool | Price from | Free plan | Runs on | Rating |
|---|---|---|---|---|
| Quote | No | Web | 3.9 | |
| Quote | No | Web | 3.8 | |
| $699/mo | Trial | Web | 3.8 | |
| $200/mo | Yes | Web | 3.6 |
Proprietary
Arkose Labs does not publish prices; all products are sold on enterprise contracts after a demo. Expect a budget for a large platform rather than a small app.
No free plan or public trial is offered. One G2 reviewer also asks for a way to test it in non-production environments, a feature the reviewer says is missing today.
No. Arkose says Bot Manager separates good users from malicious bots without friction, so each session is judged from its risk signals first. Only suspect traffic gets a challenge, and Arkose says those challenges evolve in real time against new attacks.
It identifies AI agents that visit your login, sign-up or checkout and labels each as self-disclosing good, non-disclosing good or malicious. You then set a policy per agent and per step, choosing allow, monitor or block.
Yes, SMS toll fraud is one of its named solutions, alongside account takeover and fake account creation. Arkose publishes a Device ID case study on daily savings from stopping it.
Arkose Labs is a fraud and bot defence company headquartered in the San Francisco Bay Area, with offices in Brisbane, Pune, San Jose (Costa Rica), Buenos Aires, London and Madrid and a security operations team that runs around the clock. Its platform, Arkose Titan, protects the sign-up, sign-in and in-app steps of large consumer platforms, and Adobe is among its published case studies.
The mechanism is to raise the attacker's cost. Bot Manager combines many per-session risk signals with the Arkose Global Intelligence Network to judge each visit, lets genuine users through untouched, and sends suspect traffic dynamic challenges that change as attackers adapt, so a farm of bots or paid human solvers burns time and money on every attempt. Device ID, email intelligence and phishing and scraping protection sit on the same session data. The newest layer, Agent Trust Manager, classifies AI agents such as shopping assistants into self-disclosing good, non-disclosing good and malicious, and lets the business allow, monitor or block each group at each step.
It fits security and trust teams at large consumer platforms whose losses come from fake account creation, credential stuffing that leads to account takeover, and SMS toll fraud, where automated requests run up paid verification texts. Arkose's Q3 2026 report says 97.7% of legitimate AI agent sessions carried no verified disclosure, which is the problem Agent Trust Manager is built for.
Arkose does not score payments or pay chargebacks; pair it with Stripe Radar or Signifyd for that. Teams that want one score across accounts and payments may prefer Sift, and smaller apps that need self-serve bot and takeover protection at a published price can look at Castle. See the AI security tools list.
Product facts on this page (pricing, plans, integrations, compliance) are read from Arkose Labs's own pages and documentation. The verdict is Hokai's reading of them, not the vendor's claim.