AIThis post was created with the assistance of artificial intelligence (AI).

TL;DR

Ukraine’s Ministry of Defense is using Avengers Labs to give selected defense companies controlled access to real combat drone imagery for AI training. The data stays inside a protected Dataroom, while Ukraine receives the improved models produced by participating firms.

Ukraine’s Ministry of Defense is offering selected defense companies access to a protected repository of annotated battlefield drone footage through Avengers Labs, a Brave1-linked partnership platform that lets firms train AI models on real combat imagery while Ukraine keeps the finished models.

Buying for a business?Offer from Amazon

Get business pricing on tech for your team

  • Business-only prices and quantity discounts
  • Tax-exempt purchasing
  • Multiple users, one account, clear invoices
As an affiliate, we earn on qualifying purchases.

The program centers on the Brave1 Dataroom, a protected environment that Ukraine says contains millions of annotated frames from tens of thousands of drone missions. The material includes visual and thermal imagery of aerial and ground targets under battlefield conditions, including night operations, fog, rain, camouflage, electronic warfare and multiple sensor types.

According to the Defense Ministry-linked description of the project, companies do not receive raw combat footage. They work inside the secure environment to train, validate and fine-tune models, and the improved systems are returned to Ukraine. More than 100 Ukrainian companies are described as having access, with foreign firms brought in through Avengers Labs.

The Dataroom was built by Ukrainian defense and digital agencies with the Armed Forces, a military-intelligence research institute and Palantir, according to the source material. The operational Avengers platform already uses computer vision to detect, classify and track hostile targets from drone and fixed-camera video, and Ukraine says it feeds the VEZHA module inside the DELTA battlefield-management system.

AI Dispatch · Defense

Avengers Labs

Ukraine’s Ministry of Defense is renting access to the world’s only large-scale, real-war computer-vision dataset. The terms: train your model inside the protected Dataroom — Ukraine keeps the finished AI.

0
Ukrainian companies with Dataroom access
0
Enemy units auto-detected by Avengers
Millions
Annotated frames from real drone sorties
0
Of a Shahed interception automated
01 · CAPTURE
Combat footage
Drone & camera video from the front line
▸
02 · LABEL
Annotated frames
Visual + thermal, all conditions
▸
03 · SECURE
Brave1 Dataroom
Protected env · built with Palantir
▸
04 · TRAIN
Partner models
100+ firms, Ukrainian & allied
▸
05 · RETURN
Finished AI
Improved model handed back to Kyiv
↩ The data never leaves the room. The capability flows back to Ukraine.

Inside the Dataroom

  • Structured visual & thermal imagery of aerial and ground targets
  • Hard cases: camouflaged armor, night, fog, rain, multiple sensors
  • Feeds the Avengers platform inside the DELTA / VEZHA system
  • Focus track: automatic detection & interception of enemy drones

The goal

  • 100% of frontline drones with onboard machine vision
  • Autonomous navigation in GPS-denied / jammed (EW) skies
  • Autonomous Shahed interception — human keeps the trigger
  • Scaling vs. Shahed launches rising ~35% / month
Sources: Ukraine Ministry of Defense & Min. Fedorov; Reuters, Kyiv Post, Kyiv Independent, Ukrinform, UNITED24 (Mar–Jun 2026). Weekly-detection figure per MoD reporting.
thorstenmeyerai.com

Combat Data Becomes Leverage

The program matters because combat-tested machine-vision data is difficult to replace. For defense AI, clean training imagery is less useful if models fail when targets are partly hidden, filmed at night, viewed through thermal sensors or operating in jammed skies. Ukraine’s claim is that its dataset reflects those conditions because it was captured during real missions.

The arrangement also changes the exchange between Ukraine and defense technology firms. Instead of only buying systems or hosting tests, Kyiv is offering access to a scarce training asset and asking for capability in return. That could help Ukraine improve drone detection, autonomous navigation in GPS-denied conditions and lower-cost interception of Shahed-type attack drones.

Products Worth Considering

Anker SOLIX C1000 Gen 2 Portable Power Station with 100W Solar Panel

Anker SOLIX C1000 Gen 2 Portable Power Station with 100W Solar Panel

As an affiliate, we earn on qualifying purchases.

From Delta To Dataroom

Ukraine has spent years building digital battlefield systems under wartime pressure. DELTA is its situational-awareness and battlefield-management system, while VEZHA handles video streams. Avengers is the computer-vision layer described as helping operators spot enemy equipment more quickly from drone and surveillance feeds.

The Dataroom effort follows earlier Ukrainian programs that opened the country’s battlefield to defense developers, including Brave1’s work with domestic and allied companies. A January 2026 report by The Washington Post described a Palantir-linked AI Dataroom project aimed at training air-defense systems on wartime sensor data. In May, Business Insider reported that Western defense companies were increasingly seeking to test drones, electronic warfare tools, robots and AI systems in Ukraine.

"We have a clear plan about how to stop Russia in our skies."

— Mykhailo Fedorov, Ukraine’s defense minister, reported by The Washington Post

Model Gains Still Need Proof

It is not yet clear which foreign firms are participating, how performance gains will be measured, or how widely the returned models are already deployed at the front. The public material also does not give enough detail to independently verify the size, labeling quality or full composition of the dataset.

Ukraine says the system is designed so sensitive footage stays inside the protected environment, but the exact safeguards, auditing process and export-control treatment for foreign partners remain partly undisclosed.

Partner Models Return To Kyiv

The next milestone is whether Avengers Labs can turn controlled data access into deployed battlefield improvements. Ukrainian officials are focused on machine vision for frontline drones, operation in electronic-warfare conditions and automated detection or interception of enemy drones, with humans retaining weapons-release authority.

Key Questions

What is Avengers Labs?

Avengers Labs is a partnership platform tied to Ukraine’s Brave1 defense-innovation cluster. It gives selected companies controlled access to combat data for AI training, with the resulting models returned to Ukraine.

Do companies receive Ukraine’s raw combat footage?

No, according to the project description. Companies work inside the protected Dataroom, and the sensitive data is not meant to leave that environment.

Why is the data valuable?

It shows real battlefield conditions that are hard to recreate: camouflage, night imagery, thermal signatures, poor weather, drone video, fixed-camera feeds and electronic-warfare disruption.

Is Avengers Labs already changing the battlefield?

Ukraine says the related Avengers platform already supports target detection through DELTA and VEZHA. The effect of the new partnership model on deployed AI systems is still developing.

Source: Thorsten Meyer AI

You May Also Like

Google On AI’s Distant Self-Improvement Goal—and Why Experts Like Hinton Are Growing Impatient

A headline reports Google sees AI self-improvement as distant, while experts including Geoffrey Hinton warn about risks. Key details remain unavailable.

Claude-real-video - Any LLM Can Watch A Video

Researchers develop Claude-Real-Video, a system allowing large language models to process and analyze video content directly, expanding AI capabilities.

The New Personal Agent Layer

OpenClaw and Hermes introduce a new layer of persistent personal action agents, transforming how AI interacts with digital environments.

Despite Diligence, AI Sometimes Misses The Target

Even thorough AI systems like Opus 4.8 often recognize problems but fail to complete decisive business actions, highlighting a key gap in automation.