Helsing's Clarity AI recommends strike targets — with a human name on every call

German defense company Helsing has unveiled Clarity, an AI platform that processes sensor feeds and intelligence reports into target recommendations for weapons systems, and it promises that every strike decision will be documented with the name of the human who authorized it. The company presented the platform in Berlin on Wednesday, calling it an "agentic intelligence platform." The launch puts Helsing in direct competition with Palantir, whose Maven system anchors US and NATO targeting workflows — and it arrives as operational evidence from Iran, Gaza and a near-miss with China shows exactly where such promises tend to break down.

What does Clarity actually do?

According to Helsing's press release, the software ingests all data a military unit can access: sensor feeds, open-source intelligence, written reports and systems already in use. Hundreds of specialized AI agents link those 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.

Helsing says it trained those agents on real operational data, synthetic AI-generated data and simulated environments. A shared military ontology (a common formal vocabulary for classifying military entities) and evidence base lets the agents run in parallel and deploy analytical tools. The company describes the platform as "AI-native," meaning it was not built on legacy software architectures.

The design is explicitly aimed at European buyers worried about dependence on American vendors:

• Clarity runs on the customer's own infrastructure.
• Its models and interfaces were developed in Europe, with data and operations under the armed forces' sole control.
• The platform follows the Modular Open Systems Approach (a standard for interchangeable defense components), implements NATO standards, and can connect to effectors from any manufacturer.

The stakes are straightforward: Europe wants a domestic alternative in a market currently led by US software, and Germany's Bundeswehr plans to introduce AI-powered systems starting next year.

Who signs off on a strike?

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

That accountability chain is the company's central claim. But Helsing did not say whether Clarity is already running with any customer — a notable omission for a platform pitched on operational maturity.

What happened when similar systems were used?

The record of comparable AI targeting systems is where the promises meet practice. Three documented cases frame the risk.

Iran. In the Iran war, the US military is reportedly using Palantir's Maven together 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. The paper also reports that military officials warn of an internal attitude summed up as "The computer said to do this."

Gaza. 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. A separate system called "Where's Daddy?" tracked flagged individuals to their family homes, which were then often struck at night.

China near-miss. In spring 2026, according to CNN, the US military nearly boarded a Chinese ship after a chatbot falsely claimed in an intelligence report that the vessel was carrying nuclear weapons program components. The system had combined open and classified data and misidentified the cargo. Armed troops were preparing to board and aircraft were already in the air when the error was caught shortly before the operation.

Helsing's multi-source confidence requirement is designed to prevent exactly that kind of failure. The 2026 incident shows that data fusion alone does not do it: the chatbot in that case had already combined open and classified sources before getting the cargo wrong. The lesson in plain words: redundancy helps, but it does not eliminate AI errors.

Why does the human-in-the-loop promise erode?

The pattern across all three cases is compression of human review under operational pressure. A nominal human check becomes a 20-second glance. A recommendation becomes an instruction. CNN reported that at the time of the China near-miss there were no uniform standards for verifying AI-generated information, and that younger analysts in particular trusted the tools uncritically.

"AI allows you to get to a bad idea faster," one source told CNN.

That is the structural risk Helsing's named-author rule has to survive. Documentation creates a record after the fact; it does not, by itself, slow down a decision loop that automation accelerates. Helsing's promise is that a person's name on every decision will keep responsibility anchored — but naming a reviewer does not guarantee the reviewer had time, evidence or training to disagree with the machine. The gap between the design document and the 20-second review is the gap Helsing must close in the field.

Where does this leave the European market?

Helsing enters the contest with one credential most European competitors lack: operational experience from the war in Ukraine, where the company supplies drones. Palantir's Maven, by contrast, is embedded in Pentagon and NATO workflows and has been refined through years of operational use — an advantage in a domain where field data drives capability.

Helsing's sovereignty pitch is well timed regardless. European governments have signaled demand for defense AI under national control, and Germany's Bundeswehr timeline starts next year. Whether Clarity's accountability guarantees hold up will depend less on Berlin launch materials than on the first time a commander faces a machine-generated target under time pressure.