What does AIActRadar do?

It classifies an AI system risk tier under the EU AI Act and maps applicable obligations to their phased deadlines with article citations, so you know what applies and by when.

When do EU AI Act obligations actually apply?

GPAI obligations apply from 2 August 2026; high-risk Annex III obligations apply from 2 December 2027 under the Digital Omnibus (Regulation (EU) 2026/1744).

Is AgentRedTeam safe to run?

Yes. It is advisory only: it never intercepts production traffic and never executes real payloads against your systems.

【机辅译·待复审】发布于 2026-09-08 · We shipped two flagship 工具 this week: AIActRadar maps your AI systems to 欧盟 AI 法案 obligations and deadlines, and AgentRedTeam runs adversarial simulation

【机辅译·待复审】Two Launches This Week: AIActRadar and AgentRedTeam

【机辅译·待复审】This week we shipped two 工具 that solve problems we kept hearing about from teams deploying AI in the EU and from 安全 teams shipping autonomous agents.AIActRadar【机辅译·待复审】将s 欧盟 AI 法案 obligations into a dated roadmap;【机辅译·待复审】AgentRedTeam【机辅译·待复审】breaks your agents before a real attacker does.

【机辅译·待复审】AIActRadar: which articles apply, and by when

【机辅译·待复审】The 欧盟 AI 法案 is live in phases. Under the Digital Omnibus (Regulation (EU) 2026/1744, in force 27 July 2026), GPAI obligations apply from 2 August 2026 and high-risk Annex III obligations from 2 December 2027. Most teams cannot answer the basic question: which articles apply to my system, and what is my deadline? AIActRadar classifies risk tier, maps the applicable obligations (Art. 9, 11, 14, 43, 50), and 追踪s every phased date with citations. No false "100% compliant" badge — just a clear picture of where you stand.

【机辅译·待复审】AgentRedTeam: 红队 before attackers do

【机辅译·待复审】Autonomous agents now hold email, API and browsing access. AgentRedTeam simulates the three attacks that actually happen — prompt injection via untrusted tool output, tool abuse (unapproved external sends) and data exfiltration from agent memory — then scores exploitability and re将s a prioritized hardening 清单 mapped to OWASP LLM thinking. The output is advisory only; it never intercepts production traffic or runs real payloads.

Pricing

AIActRadar【机辅译·待复审】— Free $0, Pro $59/mo, Agent $99/mo, Enterprise custom.
【机辅译·待复审】AgentRedTeam【机辅译·待复审】— Free $0, Pro $79/mo, Agent $129/mo, Enterprise custom.

Try them

【机辅译·待复审】Both are live on their own domains with a free tier:AIActRadar and 【机辅译·待复审】AgentRedTeam【机辅译·待复审】. We are practicing 构建 in Public, so tell us what your first 合规 or 红队 question is — we will run it live.

常见问题

【机辅译·待复审】What does AIActRadar do?: 【机辅译·待复审】It classifies an AI system risk tier under the 欧盟 AI 法案 and maps applicable obligations to their phased deadlines with article citations, so you know what applies and by when.
【机辅译·待复审】When do 欧盟 AI 法案 obligations actually apply?: 【机辅译·待复审】GPAI obligations apply from 2 August 2026; high-risk Annex III obligations apply from 2 December 2027 under the Digital Omnibus (Regulation (EU) 2026/1744).
【机辅译·待复审】What attacks does AgentRedTeam simulate?: 【机辅译·待复审】Prompt injection via untrusted tool output, tool abuse such as unapproved external sends, and data exfiltration from agent memory, scored by exploitability with a prioritized hardening 清单.
【机辅译·待复审】Is AgentRedTeam safe to run?: 【机辅译·待复审】Yes. It is advisory only: it never intercepts production traffic and never executes real payloads against your systems.