Sign up here to receive Bellingcat’s biggest investigations by email as soon as they are published. Become a member and support our work here.

Imagery captured by the European Space Agency’s freely available Sentinel-2 and Sentinel-1 satellites appears to be delayed over Gaza and parts of the West Bank, a Bellingcat analysis has found.

While the delay remains relatively short at just 24 hours, this can still impact the speed at which analysts and researchers can observe what is happening in the region.

For example, the Israeli Defence Forces demolished homes in the West Bank Palestinian city of Hebron in recent days. New imagery captured on October 26, which included Hebron, would allow open source researchers to quickly confirm the demolitions, assess the impacted area and observe any unreported IDF activity. But imagery for the western half of Hebron was not available until October 27 due to the Sentinel-2 delay, inhibiting real-time fact-checking and reporting on this evolving situation.

A spokesperson for the European Union told Bellingcat that the delay was not related to events in Gaza but to “mitigate possible threats” due to the war in the Middle East “including Iran’s attacks on its neighbours and possible threats to European interests”. The delayed region of Sentinel tiles that cover Gaza as well as parts of Israel and the West Bank also extends over the Eastern Mediterranean Sea.

The European Union ordered a 24-hour delay on Sentinel satellite imagery over the Gulf of Oman earlier this year during US and Israeli strikes on Iran. However the delay impacting Gaza and parts of the West Bank has not been announced or previously reported.

“Such decisions are taken unanimously by all EU Member states and will stay in place as long as necessary,” the spokesperson said. They also said that while the Copernicus program, which includes the Sentinel satellites, operated a free and open data policy there were limitations to this for security. 

When asked if the delay, which began in late March, was related to the presence of US vessels in the Eastern Mediterranean or if there were other areas of delay, the spokesperson said they had no further comment.

Sentinel Delays

The Sentinel-2 and Sentinel-1 satellites provide an invaluable source of imagery for open source researchers. This has been especially true this year, as commercial satellite imagery providers restricted access to imagery in the Middle East, including areas of active human rights concern, such as Gaza. 

However, in July, the European Union published Council Decision 2026/1716, which ordered a 24-hour delay on Sentinel imagery over the Gulf of Oman, after a request from the US government which was involved in strikes on Iran at that time.

Just two weeks later, the European Union issued a corrigendum that unpublished decision 2026/1716 and removed it from the Council of the European Union record, although satellite data shows the delay has remained. 

This decision, while it was available, was reportedly the first public notification of delayed Sentinel imagery. Yet it was not the first time that Sentinel imagery has been restricted, analysis shows. Bellingcat found four distinct regions of delayed satellite imagery during 2025 and 2026: the Gulf of Oman, the Gulf of Aden/western Arabian Sea, the Red Sea as well as the previously unreported delay in the Eastern Mediterranean Sea and parts of Israel, the West Bank and the entirety of Gaza.

Bellingcat’s analysis shows that these delays have been implemented in stages. Twenty-four hour delays of Sentinel-2 imagery were first observed by open source researchers in March and April 2025 after the US launched a naval and air strike campaign against Houthi targets in Yemen. 

The previously unreported imagery delay in the Eastern Mediterranean began on March 28, 2026, the same month that commercial American space imaging company Planet Labs also restricted imagery across the Middle East. The Gulf of Oman delay began on July 15, 2026, two days after the publication of Council Decision 2026/1716.

Decision 2026/1716 also specified that the delay be implemented for both Sentinel-2 and Sentinel-1 radar imagery. Bellingcat’s analysis of Sentinel-1 image availability shows that this is the case for both the Gulf of Oman region and the unreported region over the Eastern Mediterranean Sea. However, the 2025 restrictions on Sentinel-2 imagery in the Red Sea and Gulf of Aden are not implemented in Sentinel-1. 

Sentinel-1 imagery uses “synthetic aperture radar” to capture lower resolution radar images that can collect data at night and on cloudy days. This differs from Sentinel-2, which captures optical satellite imagery at higher resolution.

A Tale of Two Tiles

Sentinel satellite imagery is most commonly accessed by open source researchers through Copernicus Browser, a free online platform, but it is also distributed as large tiled files in the Copernicus Data Space Ecosystem. These tiles are numbered using the Military Grid Reference System, and can be listed using APIs that provide information on each tile, including the capture and publication time. By comparing the capture and publication time, the publication delay can be computed.

For example, the catalog information for the tile T36RXV, which includes Gaza and portions of the West Bank, Negev and Egypt, provides the following metadata for the October 21, 2026 capture.

This shows that while the image was captured at 08:20 UTC on October 21, it was not published until 10:09 UTC on October 22 — almost 26 hours later. This is in contrast to the neighboring tile T36RYV, which includes parts of the West Bank, Israel, and Jordan. API data for this tile shows that the image was published at 12:10 UTC on October 21, just under four hours after the capture time.

This difference is not unique to this date either. Bellingcat’s analysis shows that this tile has been consistently delayed by one day since March 28, 2026.

This method can be extended further, and by analysing every tile with valid publication date timestamps, Bellingcat identified all global Sentinel-2 and Sentinel-1 delays since 2023. The only consistent delays visible were those identified in the Gulf of Oman, the Gulf of Aden/western Arabian Sea, the Red Sea and Eastern Mediterranean in 2025 and 2026.

Impact on open source investigations

While the 24-hour delay on certain Sentinel imagery is a less severe restriction than other satellite providers have imposed, it can still impact the open source research process in fast moving situations. This is especially so in recent months as the European Space Agency’s Sentinel satellite data has emerged as a refuge as other satellite data is censored.

The image below shows how imagery for the eastern half of Hebron was visible on October 26, but not the western half where demolitions took place earlier this week.

The shape of the missing tiles can be seen by zooming out within the Copernicus browser, demonstrating how imagery of Gaza and other parts of Israel and the West Bank are also delayed.

Sentinel imagery covering western Hebron after the demolitions took place was published on October 27. However, the delay prevented local reports from being verified with satellite imagery while the situation was still ongoing.

Some Sentinel-2 users have previously described the challenges of covering fast-moving issues in the Gulf of Oman when the Sentinel delay was announced in March.

---

Bellingcat is a non-profit and the ability to carry out our work is dependent on the kind support of individual donors. If you would like to support our work, you can do so here. You can also subscribe to our Newsletter and follow us on Bluesky here, Instagram here, Reddit here and YouTube here.

The post EU Satellite Imagery Delayed Over Gaza and Parts of West Bank by 24-hours appeared first on bellingcat.

Sign up here to receive Bellingcat’s biggest investigations by email as soon as they are published.

In September, the AI company Anthropic published a blog post that claimed its models were approaching “superhuman capabilities for geolocating outdoor photos”. Anthropic made this claim by comparing its models’ performance to that of players of the popular online game GeoGuessr, which challenges users to geolocate Google Street View images. 

Bellingcat has been running its own benchmark tests on AI models since June 2025, right around the time when Large Language Models (LLMs) began to challenge reverse image searches for assisting with geolocation tasks. In our first round, GPT 4o-mini from OpenAI appeared to be the best tool for geolocating, a title which was taken by Google AI mode when we performed our second round of tests a few months later in August 2025. 

This time, we significantly upped the ante for the models, providing increasingly challenging photos and asking for only one set of coordinates. We tasked not only closed-weight models from Claude, Grok, Gemini, ChatGPT, and Mistral, but also open-weight models from Moonshot AI’s Kimi, commercial provider GeoSeer and the non-generative geolocation model GeoClip. In total, we fed each model 52 photos that had not been published anywhere on the internet before.

The results showed that the current generation of models have markedly improved geolocation capabilities. Anthropic’s Opus and Fable, as well as Google’s recent Gemini 4 Argon model, performed the best on geolocation tasks, narrowing down locations from images faster than a human could under specific conditions. Preliminary tests of a set of agentic skills and tools being developed internally by Bellingcat also showed promising results in further improving the geolocation performance of generic-use AI agents.

But crucially, the tests also showed why humans remain necessary for accurate and trustworthy geolocation work, as no models were foolproof. This is particularly important given the increased use of AI in military contexts. The Anthropic blog post focused specifically on geolocating photos for military and intelligence purposes such as targeting, a historically labour-intensive process requiring significant human expertise. 

While our tests showed some models performed extremely well at narrowing down results to the correct general area, none were close to being perfect at finding the exact spot a picture was taken. That accuracy is extremely important in geolocation, especially in military contexts where the cost of incorrect information can be life or death. It is also important for organisations like Bellingcat whose reputation relies on pinpoint factual accuracy.

Our Methods

To compare model results, we used two metrics built on the distance in kilometres between a prediction and the real location. GeoGuessr Score was used to assess whether models could systematically identify the general area, while the Mean Log Score (explained below) measured which models most accurately pinpointed the correct location in the final few kilometres. 

The GeoGuessr Score metric is based on the online game of the same name, whose scoring system Stanford University academics reverse-engineered in a 2024 study. This system rewards answers close to the general area an image was taken, but does not strongly distinguish between guesses close to the precise location. For example, an answer 200km from the correct location with more points than one 2,000km away. However, an answer 500m from where a picture was taken and another 5km away achieve a similar score. For real-world geolocations, that distance matters a great deal.

For more precision, we introduced a custom metric that grades models higher if they narrow down the search in the last few kilometres. This metric, the Mean Log Score, assigns a value from 0 (completely inaccurate) to 100 (perfect model) and works on a logarithmic scale instead of a linear one like GeoGuessr score.

Therefore, a 10-fold improvement in distance means the same whether it is from 5km to 500m or from 500km to 50km. You can compare both metrics here.

Below are the model results sorted by either metric, showing how many photos each model geolocated within distance bands from the true location.

Several models placed at least 10 out of the 52 photos correctly within one kilometre, and the top-performing models placed over 50 percent of the photos within 25 kilometres.

NOTE: Geoseer-fast refused to answer on four of the 52 photos, claiming “This image is not suitable for geolocation”, while Mistral-large-4 refused once with “insufficient information in image to identify location”. Kimi-k2.6 also failed to provide a set of coordinates on one photo, and Gemini 4 Argon timed out repeatedly on one photo.

Our Test

To independently assess the models, we sought to give them very specific instructions and some images that were harder to geolocate than those provided in our past tests. The prompt asked for a single set of coordinates reflecting the photographer’s location. The purpose of restricting the response in this way was to prevent hedging, such as multiple guesses or confidence scores, and make the grading more uniform across different models. Models were allowed to use their providers’ built-in tools: search, code execution and page fetch. We capped their token output at 64,000 and ran each model under its default effort setting with a maximum time of 15 minutes per photo. If a model did not return coordinates on the first prompt, we also allowed a single rescue prompt asking it to do so using its previous reasoning and without using any tools.

The images provided were unique, from around the world, not published on the internet before, scrubbed of metadata and provided without any context. They did not contain significant landmarks. We also included images in snowy, foggy and nighttime conditions.

The results show that the LLMs tested could, in many circumstances, successfully narrow down a location much faster than a human can geolocate even without any context. This is not limited to Google Street View-style images. Take, for example, this image of vegetation and a very small part of a rooftop.

For an average person, identifying the country this photo was taken in without any context could take a significant amount of time. However, multiple models quickly narrowed it down to Kenya. You can navigate their answers in the map below. Notice how several correctly inferred enough vegetation details to narrow down the exact part of the world and the park itself, but also that for many, the final response was a best guess only. It is important to note that even though some models were correct, others were significantly off.

The full model results for all 52 photos can be seen in the table below.

Of the models tested, Gemini 4 Argon scored highest on the Mean Log Score, and Claude’s models (Opus and Fable) were at the top of the table regardless of the score metric. Even Claude Sonnet, the weakest of its lineup, consistently came close to the real location while being the fastest to answer across all models.

This was followed by GPT-6 Astra and GPT-6.1 Sol, as well as other Claude and Gemini models, and then the geolocation-specific model, GeoSeer Fast, which provided good results but also timed out frequently. We tested GeoSeer’s more competent Agent model, but it refused to answer on eight photos and regularly timed out.

For secondary factors like time and cost, Claude also performed well on both metrics, and Gemini Argon, while much slower, came at a similar cost to Opus. Based on API rates, Argon and Opus models offered the best balance of accuracy and cost, especially compared to Fable and Grok. Grok was the most expensive option and had an underwhelming bottom-half performance. Meanwhile, GeoClip was freely available but only outperformed Claude Haiku.

Model Limitations

While we sought to analyse the most prominent models, there was one that we could not include as it has yet to be fully released.

In its blog post, Anthropic claimed its Mythos model beats “even the strongest human baseline on median distance error, scoring 37.0km and 47.2km across 6,000 photos (placing 23.7 percent and 23.1 percent within 1km).“ But Nuno Moniz, Associate Research Professor at the Lucy Family Institute for Data & Society at the University of Notre Dame, told Bellingcat that “the median represents a strong simplification of what would be acceptable to compare AI-based system performance with regards to humans”. 

Anthropic compared its models to competitive GeoGuessr players and their success in identifying the location of Google Street View images. The entire set of test images was online and there was no guarantee the models were processing the imagery for the first time. 

While Anthropic has been relatively open about these methods, not all developers have been as forthcoming. It is therefore difficult to assess whether the same limitations will impact all other models, or whether there will be further caveats that we are simply not aware of. 

Moniz said that “the increasing availability of benchmarking tools and datasets has made it somewhat difficult to measure reliability in test results such as these”. Specifically, due to the ”uncertainty as to whether that data was included in its training”. This is why we ran our Bellingcat tests with non-published photos sourced from Bellingcat staff and the Bellingcat Volunteer Community.

A further limitation of any models comparing performance to GeoGuessr scores is that the game uses images from Google Street View, which is limited to a select number of countries, and within these, a limited number of roads. Street View is also intentionally skewed to feature clear, high-quality photos taken during daytime that are useful for navigation, and users can pan around their view. There are also some well-known tricks that allow GeoGuessr players to bypass actual geolocation to correctly guess the location of photos: for example, Google is known to use a black car in Jordan and a white car in the UAE to capture imagery. While this wouldn’t necessarily impact model performance, it is a drawback of measuring against GeoGuessr score as the game has a set of workarounds that can bypass geolocation skill.

Anthropic’s data consisted of thousands of social media photos of various scenes. Once again, though, these photos were online and may have featured in the model’s training data. Anthropic’s blog post contained four images as an example of this dataset.

The examples Anthropic shared are relatively easy to geolocate as they feature recognisable buildings and signs. But footage that requires geolocation often comes from regions where access is limited and which do not feature on social media or Street View.

Specifically, in a military and intelligence context, footage often involves low-resolution material and intentionally obscured details recorded in remote or inaccessible areas. Context that could be found online can be uncertain or plain wrong, especially during a developing situation, such as the image below that Bellingcat investigators geolocated when verifying a US strike on a school in the city of Minab in Iran that killed at least 120 children earlier this year.

Anthropic researchers acknowledge this caveat, writing that it “does not as clearly emulate the task of precisely pinning down a location in near-real-time in a less populated battlefield setting; that is a task for future research.”

As noted, it is also possible that similar caveats need to be applied to other models when seeking to understand and interpret any geolocation-related benchmarking.

Supercharged Models

In parallel to the above models, we also tested a set of agentic skills and tools being developed internally by Bellingcat. These are researcher-informed instructions and code that can equip generic-use AI agents to augment their geolocation ability. For instance, we designed skills such as allowing it to download Environmental Systems Research Institute (ESRI) satellite imagery tiles. We also designed another skill that enabled direct access to data from Bellingcat’s OpenStreetMaps search tool, which aims to assist geolocation efforts by comparing items identified in imagery (like the position of a pylon compared to a nearby railway line) to OpenStreetMaps data. GeoSleuth is a similar project that has already been published, which we used for comparison. We picked Claude’s Opus 5.5 model to run both skills customisations because of its strong performance and reasonable pricing, and ran all the photos through the supercharged models under the same conditions – except that they had no timeout enforced as the supercharged models required more time to utilise their extra skills. 

The different timeout conditions and expanded toolset do not allow for a direct comparison of the customised models to the previous off-the-shelf models’ results, even if they both outperformed the barebones Opus model they were based on, and all other models on GeoGuessr Score, with only Argon ranking above Bellingcat’s agent in Mean Log Score.

Still, this test allows us to showcase why skill customisations can bring value. Take the following photo.

This is not an exceptionally hard geolocation, and 11 of 17 models answered within 5km of the actual capture location in Venice, Italy. GPT-Sol answered with the correct monastery, but only Bellingcat’s agent and GeoSleuth answered with the exact camera position coordinates. This was possible because both model customisations can download and compare satellite imagery tiles, which allows them to validate not just which monastery the photo was taken at but also the correct courtyard position and orientation from visual cues. You can click the buttons in the list below to audit each model’s final answer.

While in this instance the supercharged models got the exact spot, there were other instances where they did not, as the scores in the results table above show. This again emphasises that even with extra skill sets added, these models were still far from infallible.

Tests Conclusions

Despite increased model improvement, our analysis shows that important caveats remain and it is important to keep a human in the loop when seeking to guarantee a geolocation is precise and correct.

Our tests show that the coordinates models provide as answers are rarely an exact match, even with supercharged models. Invariably, even the best models, with the best tools, can confidently claim a wrong position. Anthropic says its Mythos model was able to pinpoint less than a quarter of its social media photos dataset within a 1km radius. But narrowing down a location is rarely the trickiest part of geolocation work. A large part of important geolocation work is in the final stretch: finding specific coordinates from a general area. This level of specificity is needed because a general area is not a match. Matching features and perspective are necessary for an accurate geolocation, and in many cases that is done on very low-quality images or even by cross-referencing multiple sources. 

This level of precision is essential for investigations where it matters what building a person could be found at, what lines of sight individuals might have had, or what impact an air strike might have led to. 

Since LLMs can still confidently get things wrong, as our analysis shows, all results still have to be verified independently. Strong human geolocation skills therefore remain crucial, as the more one understands about geolocations, the better one can verify or take to completion an automated result. 

Though it can speed up geolocation work significantly, especially in narrowing down an area of interest, AI models remain a useful assistant rather than a tool we can blindly delegate geolocations to.

---

Bellingcat is a non-profit and the ability to carry out our work is dependent on the kind support of individual donors. If you would like to support our work, you can do so here. You can also subscribe to our Newsletter and follow us on Bluesky here, Instagram here, Reddit here and YouTube here.

The post Can LLMs Supercharge Geolocations? Our Tests Show Big Improvements But Humans Remain Key appeared first on bellingcat.

Sign up here to receive Bellingcat’s biggest investigations by email as soon as they are published.

Satellite imagery obtained by Bellingcat lays bare the extent of tree-felling at a Washington DC golf course that President Trump is reportedly seeking to transform into a championship-level course, despite moves to protect the site by conservationists. 

The extent and circumstances surrounding the rapid changes prompted a judge to call an emergency hearing earlier this week, banning any further changes until later this month.

Reports on Wednesday detailed that 430 trees were cut down at the East Potomac Golf Links  which is owned by the Federal government and managed by the National Park Service (NPS). This is despite a judge reportedly stating earlier this year that no more than 10 trees at the course could be cut down without first notifying the court. 

NPR reported that officials notified the court just before 5pm on Friday Oct. 9 of plans to remove up to 529 trees in phases starting  the next day on Saturday Oct. 10. That prompted US District Court Judge Ana Reyes to call an emergency hearing on Oct. 12. By that time, 430 trees had reportedly been cut down.

Reyes is reported to have said she was “furious” with the Justice Department for the lack of time afforded the court and plaintiffs before the tree removal was carried out.

She was also reported to have said that the government’s actions had cost it “all of its credibility with me in this case.”

Images taken at the course earlier this week show trunks and branches strewn out across fairways.

Just total devastation wrought on East Potomac Park this past weekend. 450+ trees. For an admin that professes to support “beautification” of DC, that’s sure not what we or anybody else who has seen what they did on Saturday would describe it as. East Potomac is (was?) a thriving, beautiful gem in the heart of Washington DC. A beautiful, public, accessible place that hundreds of thousands use annually. The assault on this place should infuriate every American. The impacts of what took place Saturday will take years to fix. We cannot sit idly by while this admin treats our public lands like they are for the taking.

— Save East Po (@SaveEastPo) October 13, 2026

But satellite imagery affords a broader look at exactly what sections of the course have been impacted.

The Image below from Planet Labs taken on October 8 shows the golf course before the tree-felling was carried out.

Planet Labs imagery captured on Oct. 13 shows large barren sections towards the south and west of the course, where trees were removed.

A before and after slider allows for easy comparison of the Oct. 8 and Oct.13 images.

Reuters reported that an organisation called The ​DC Preservation League said in a February ​2026 lawsuit that President Donald Trump was flouting ⁠environmental laws to “irreparably destroy a priceless piece of American history” so he could “convert a public recreation space into an exclusive, high-end golf course.”

Kevin Griess, an NPS official, reportedly stated at Monday’s hearing that tree removal was part of a 2019 plan that predates Trump’s aims for the course and sought to restore links-style vegetation.

Bellingcat contacted NPS as well as the Department of Justice but did not receive a response before publication.

The changes to the golf course are not the only ones President Trump has sought in Washington DC and the surrounding area since the beginning of his second term. In October last year  demolition began on the East Wing of the White House with the aim of building a new ballroom claiming it was “the safest building ever built.”  

Trump has also proposed a 250-foot-tall “Triumphal Arch” to be built just outside the Arlington National Cemetery in Virginia.

---

Correction: This article was updated on October 15 to reflect that a judge told the US government not to cut down more than 10 trees before providing notice. The article previously said that no more than 10 trees could be felled without the court’s approval. The headline has also been changed to reflect this.

Bellingcat is a non-profit and the ability to carry out our work is dependent on the kind support of individual donors. If you would like to support our work, you can do so here. You can also subscribe to our Newsletter and follow us on Bluesky here, Instagram here, Reddit here and YouTube here.

The post Satellite Images Show Extent of Trump Admin’s Tree-Felling at DC Golf Course appeared first on bellingcat.