Helsing 发布 Clarity 智能体平台，为武器系统筛选打击目标

The Decoder：AI News· Maximilian Schreiner·2026-11-11 23:12· 47 分钟前AI 评分49

Helsing built an AI system that prepares targets for weapons systems and says every decision will carry a human name

AI 导读

德国国防科技公司 Helsing 在柏林发布 agentic 智能平台 Clarity，用数百个专用 AI 智能体融合传感器、开源情报与报告数据，识别军事目标并生成打击建议，最终由人类选定目标并授权移交武器系统。

正文 · 原文

German defense tech company Helsing has unveiled Clarity, an agent-based software platform that distills sensor data into target recommendations. Helsing promises European control and human oversight. The use of comparable systems in Iran, in Gaza, and in a near-miss involving China shows where those promises break down.

Helsing presented its Clarity platform in Berlin on Wednesday, calling it an "agentic intelligence platform." According to the company's press release, the software processes all data a military unit can access, including sensor feeds, open-source intelligence, written reports, and systems already in use. Hundreds of specialized AI agents link these sources, identify military forces, facilities, and positions, and prepare information on adversary assets for human review before Clarity packages the data for the appropriate weapons system.

The agents are trained on real operational data, synthetic AI-generated data, and simulated environments, Helsing says, and a shared military ontology and evidence base lets them work in parallel and deploy analytical tools. The platform is "AI-native" and not built on legacy architectures, according to the company.

Helsing stresses the system's sovereignty, noting that Clarity runs on the customer's own infrastructure and that its models and interfaces were developed in Europe, with data and operations staying under the armed forces' sole control. The platform follows the Modular Open Systems Approach, implements NATO standards, and can connect to effectors from any manufacturer.

The decision to strike is supposed to stay with a human who selects targets and must authorize every handoff to a weapons system. A target is only recommended, Helsing says, when independent sources agree and the assessment can be made with high confidence, and every decision is documented with the name of the person responsible.

With Clarity, Helsing is stepping into a market led by Palantir, whose Maven system is embedded in Pentagon and NATO workflows and has been refined through years of operational use. Helsing is one of the few European vendors that can point to experience from the war in Ukraine, where the company supplies drones, though it didn't say whether Clarity is already running with any customer. Germany's Bundeswehr plans to introduce AI-powered systems starting next year.

The speed of automation creates its own risks

How these systems play out in practice is visible in the Iran war, where the US military is reportedly using Palantir's Maven with Anthropic's Claude to suggest hundreds of targets, prioritize them, and recommend ordnance. In the first 24 hours the US struck about 1,000 targets, and within days that number topped 3,000, according to the Wall Street Journal, which also reports that military officials warn of an attitude described internally as "The computer said to do this."

The problem is sharper in Gaza, where according to investigations by +972 Magazine and Local Call, Israel's Lavender system for a time flagged up to 37,000 Palestinians as potential targets. Human review often lasted only about "20 seconds" per target, mainly to verify whether the person was male, the investigations found, and a separate system called "Where's Daddy?" tracked flagged individuals to their family homes, which were then often struck at night.

Data fusion doesn't prevent hallucinations

Even combining multiple sources is no guarantee, as an incident from spring 2026 shows. According to CNN, the US military nearly boarded a Chinese ship because a chatbot falsely claimed in an intelligence report that it was carrying nuclear weapons program components after combining open and classified data and misidentifying the cargo.

Armed troops were preparing to board and aircraft were already in the air when the error was caught only shortly before the operation. There were no uniform standards for verifying AI-generated information, and younger analysts in particular trusted the tools uncritically. "AI allows you to get to a bad idea faster," one source said.

AI News Without the Hype – Curated by Humans

Subscribe to THE DECODER for ad-free reading, a weekly AI newsletter, our exclusive "AI Radar" frontier report six times a year, full archive access, and access to our comment section.

来源：The Decoder：AI News · the-decoder.com