Arceris Reviews & Product Details

Arceris is a self-hosted AI security gateway. It routes an organization's AI traffic through one on-premise control point and enforces policy on it before anything leaves the network. Traditional DLP was built for files and email. It does not read what an employee types into ChatGPT, the document they attach, or the answer that comes back. Arceris does. Coverage comes from routing rather than integration, so it reaches surfaces an SDK or a browser extension cannot: - Browser: ChatGPT, Claude and Gemini web sessions, via transparent HTTPS inspection with nothing installed on the client - Terminal and IDE: Claude Code, Cursor, Codex, Gemini CLI and MCP servers, through a local agent with zero-touch Kerberos enrolment - API: a unified gateway endpoint, one key per user across providers - Self-hosted models: llama.cpp, vLLM, Ollama and other on-premise model servers One policy set applies at every door: sensitive data is detected and masked (28+ patterns, including Turkish national ID and IBAN), prompt injection and unsafe content are blocked, uploaded files and images are inspected, multi-turn conversations are scored for escalation, model answers are checked for data-exfiltration markup, and each team's model access and token spend are enforced. Every decision is logged, exported to SIEM, and summarised in periodic reports mapped to the OWASP Top 10 for LLM Applications and MITRE ATLAS. Arceris runs entirely on the customer's own infrastructure and supports fully air-gapped deployment: detection models, OCR and DLP integration all run inside the perimeter. There is no Arceris server in the request path. It connects to what an organization already runs — an existing ICAP DLP server, LDAP, Kerberos, OIDC, agent-based SSO and SIEM — rather than replacing it.