Irish Minister for Foreign Affairs Helen McEntee delivers a speech at Dublin Castle in Dublin, during the opening ceremony of Ireland's EU presidency on Wednesday July 1, 2026. (Niall Carson IRISH Press Association via AP Images)

The Irish Presidency of the Council of the European Union Should Welcome Scrutiny from Academia

Illustration of a Tamagotchi with the OpenAI logo depicted on its screen. Tech Policy Press/Shutterstock

Illustration of a Tamagotchi with the OpenAI logo depicted on its screen. Tech Policy Press/

When Users Say 'Goodbye' to AI

UK Prime Minister Andy Burnham speaking during a visit to a family center in Sheffield, South Yorkshire, on Friday, July 31, 2026. (Annabel Lee-Ellis/PA Wire POLITICS)

Taking Stock of the UK Tech Policy 'Muddle'

Screenshot from a Google announcement depicting its launch of a Nano Banana 2 feature in Google Earth. 

Google Earth AI Fiasco Underscores Why Tech Firms Must Listen to Outside Experts

Mayor Zohran Kwame Mamdani and Chief Technology Officer and Office of Technology & Innovation (OTI) Commissioner Lisa Gelobter announced the launch of "Public Interest Technology (PIT) Crew," a new initiative to deploy teams of technologists alongside City agencies to tackle public problems with in-house digital solutions at Coney Island in New York City on July 13, 2026. Credit: Katie Godowski/MediaPunch /IPX

Cities and States Can Replicate NYC’s Ambitious Public Interest Tech Crew

This image is a collage with a colourful Japanese vintage landscape showing a mountain, hills, flowers and other plants and a small stream. There are 3 large black data servers placed in the bottom half of the image, with a cloud of black smoke emitting from them, partly obscuring the scenery.

Deborah Lupton, "Servers in a Landscape" / 

Is Open-Source AI Really the Dangerous Path?

Telegram on App Store displayed on a phone screen and French flag displayed on a screen in the background are seen this illustration photo taken in Krakow, Poland on December 3, 2023. 

(Photo by Jakub Porzycki/NurPhoto via AP)

What a Growing Telegram Disinformation Network Reveals About the DSA

A woman waves a European Union flag in the wind.

The ‘Sovereignty’ Narratives That Don’t Help Europe

Kherson, Ukraine—Lenka Klicperova, journalist and photographer, covers the war in Ukraine. October 19, 2022 Jose Hernandez/Shutterstock

Kherson, Ukraine—Lenka Klicperova, journalist and photographer, covers the war in Ukraine. October 19, 2022 Jose Hernandez/

What It Now Takes to Establish a Fact

Editorial collage in black, white, and blue showing a megaphone emerging from a laptop screen and a hand holding a microphone, surrounded by ribbons reading 'NEWS' and scraps of newsprint.

How Cyberattacks Became a Censorship Tool

OpenAI founder and CEO Sam Altman speaks with reporters as he departs a meeting with Sen. Raphael Warnock (D-Ga.) on Capitol Hill on July 29, 2026. (Francis Chung/POLITICO via AP Images)

The OpenAI–Hugging Face Incident Demands Urgent Congressional Oversight

Nine small images with schematic representations of differently shaped neural networks, a human hand making a different gesture is placed behind each network.

Can AI Agents Be Aligned with Human Rights?

The X logo is being displayed on a mobile phone with the European Commission logo seen in the background, in this photo illustration. Taken in Brussels, Belgium, on 8 December 2025. (Jonathan Raa / Sipa USA) (Sipa via AP Images)

X's Data Access Remedies Are a Win for Researchers Only If They Last

A person holds a Mastercard debit card near a mobile phone with a contactless POS terminal in Athens, Greece, on February 15, 2026. (Photo by Nikos Pekiaridis/NurPhoto via AP)

(Photo by Nikos Pekiaridis/NurPhoto via AP)

Pakistan’s Banking System Holds a Lesson for Social Media Age Verification

A view of the Barlaymont European Commission building in Brussels, Belgium on July 20, 2026. (Photo by Jakub Porzycki/NurPhoto via AP)

 Brussels Gains New AI Act Enforcement Powers as Autonomous AI Tests Regulators

The EU flag with an image of the X logo. 

What X's Corrected Action Under the DSA Fixes on Data Access, and What's Left Open

OpenAI founder and CEO Sam Altman is seen in an elevator after a meeting with Sen. Ted Cruz (R-Texas) on Capitol Hill on July 29, 2026. (Francis Chung/POLITICO via AP Images)

Advanced AI Is Ultrahazardous. Let’s Treat It That Way

Inside the frame of a mobile phone screen, a woman is depicted walking and looking over her shoulder. Behind her, there is an outdoor scene with the sky, tree, fences, and pavement. A hand is shown grabbing the phone and tapping the record button. On the left side of the phone, a small portion of the same sky, fences, and pavement is displayed in a different scale, representing the world outside the screen. On the right side of the phone, three figures of the same woman are shown against a plain blue background, progressively transforming from minimally glitched to heavily glitched.

Consent, Not Nudity, Should Define Image-Based Abuse Laws

Britain's Prime Minister Andy Burnham is shown the driver's area on a FirstGroup bus at the bus station during a visit to Bath, England, Wednesday July 22, 2026. (Toby Melville/Pool Photo via AP)

Burnham's First Week Redraws the Map of UK Tech Governance

A child using a laptop computer. (Press Association via AP Images)

Brazil's Child Online Safety Law Offers a Different Path Than Social Media Bans

Audio of this conversation is available via your favorite podcast service.

The United Kingdom has a new prime minister, and as Tech Policy Press fellow James Ball reported last week, he has already moved to 

—scrapping a scheme for digital ID, shuttering the Department of Science, Innovation, and Technology (DSIT), shifting the Government Digital Service (GDS) to the Department of Culture, Media, and Sports, and reshuffling responsibility for AI policy. 

To learn more about how the table is being set under Prime Minister Andy Burnham, I spoke to three individuals who are following these matters closely:

, the founder director of a new think tank called 

 to Google Earth, giving users the ability to use its Nano Banana image generation capabilities to “transform” real world locations. After public outcry, on Friday the company 

 the new feature in order to work on “implementing stronger guardrails.”

The incident is just one more parable in a long line of product releases by AI firms with potentially dangerous consequences for information integrity. In this case, Google didn't just 

. It briefly changed one of the world's most trusted sources of evidence. For more than two decades, Google Earth has earned the trust of journalists, human rights investigators, open-source researchers, and even courts. Not because it is flawless, but because it provides a stable, time-stamped record of the physical world.

That credibility has made it part of the global verification infrastructure. By integrating generative image creation, Google threatened to fundamentally 

 that relationship. Google Earth was no longer just a record of reality, but also a generator of plausible realities, collapsing the boundary between one of the world’s most trusted reference systems and a generative AI system.

This episode revealed something much bigger than a single product mistake. It showed how easily generative AI can undermine the trust embedded in public information infrastructure when governance fails to keep pace and when companies prioritize commercial and consumer priorities over reality resiliency. And it is just another point of evidence as to why companies like Google must more actively engage with the civil society experts who could have seen this coming.

At WITNESS, we warned about exactly this in our 

2024 report on generative AI and conflict

. One of the key risks we identified was the ability to generate synthetic evidence of military operations, humanitarian crises, and damaged infrastructure tied to real places, making verification during conflict substantially harder.

We have already seen how this plays out. For example, during the recent Iran-Israel-US war, 

 purporting to show destroyed US military infrastructure spread widely online. The fabricated images were built from authentic Google Earth imagery of a US naval base in Bahrain, adding convincing synthetic destruction while preserving real geographic features. They accumulated hundreds of thousands of views before BBC Verify and other investigators established they were fake. None of this should have come as a surprise. We argued about this in our December 2025 Reuters Institute article on how AI is 

 exactly how these risks emerge in practice.

The greatest risk is not a single convincing fake. It is the normalization of ambient uncertainty: an information environment where authentic evidence, synthetic media, misleading forensic analyses, and faulty AI detection results circulate together until distinguishing reality becomes increasingly difficult.

 we saw exactly that. Genuine evidence was dismissed as AI. Fake evidence was used 

 were weaponized to undermine authentic reporting. The objective was no longer to convince people of a lie; it was to make reality itself feel unknowable.

Google’s rollback should be welcomed. But pausing one feature does not resolve the deeper issue that its brief deployment exposed. Google 

 the capability with examples like creating "a rendering of your future house nestled perfectly in the actual landscape." But products should not be evaluated solely by their intended use cases. They must also be judged by their foreseeable misuse. In this case, the potential misuse cases made it obvious that meaningful consultations with media and civil society organizations would have identified them before launch.

And companies like Google also have a responsibility when they recklessly expand the Overton window on what is acceptable or normal to manipulate. It’s true that it doesn’t require a Google tool to manipulate a satellite or geo-image—and that satellite and geospatial images were photoshopped before this (as 

 in its MH 17 investigations), and also that other tools could be used to achieve similar results. Satellite imagery has been a target of concern in deepfakes analysis and 

 since the earliest days. But there is an enormous difference between something being technically possible and being seamlessly integrated into one of the world’s most trusted geographic platforms. By making this capability native, frictionless, and globally accessible, Google dramatically lowered the barrier to producing location-based fabrication while normalizing manipulation on a hitherto relatively ground-truthed source.

The problem is compounded by a verification ecosystem that is already struggling to keep up with the increasing volume and variety of synthetic content. Today's AI detection systems were largely built to identify manipulated faces and biometric signals—not synthetic environments, landscapes, damaged infrastructure, smoke, satellite imagery, or disaster scenes. As

, the verification ecosystem is fundamentally mismatched to the kinds of deceptive content that increasingly matter in conflict, climate disasters, and humanitarian crises.

This is also a verification ecosystem where plausible deniability is routinely deployed to dismiss real content as artificially generated. The history of satellite imagery in human rights work is of its use to prove war crimes in 

; competing with falsified images from the ground that claimed to prove events didn’t happen or that 

 Now, the satellite imagery can be generated to reinforce the lie, and the ‘liar’s dividend’ deployed to dismiss what has been a hitherto trusted source.

Perhaps most striking is how casually this capability was introduced. Google's initial response emphasized safeguards largely focused on downstream mitigations such as SynthID watermarking and content restrictions. Those are useful measures, but they are not an adequate governance strategy. Products that become part of the world’s verification infrastructure require fundamentally different standards for deployment––this is a feature that reshapes one of the world's most trusted sources of visual evidence, and that places yet another assumption of ‘must verify everything you see on the internet’ upon us all.

Platform capabilities with this level of societal impact demand rigorous pre-deployment risk assessments, meaningful consultation with affected communities and civil society experts, and safeguards proportionate to the risks. Watermarking is an important mitigation, but it is not a substitute for responsible product design. In the real world, images are screenshotted, cropped, compressed, reposted, and stripped of provenance within minutes. They leave Google's ecosystem almost immediately, while the burden of sorting truth from fiction falls on journalists and investigators using verification tools that already struggle to analyze precisely this type of content.

That is exactly backwards. The burden should not fall on verification communities to absorb the consequences of increasingly powerful generative systems while tech firms profit. Companies introducing technologies that reshape the evidentiary landscape should have to demonstrate that they can be deployed safely before they become globally available, not ask the rest of the information ecosystem to adapt after the fact. Google’s decision to pause the feature suggests those concerns are now being taken seriously. The question is why they were not central to the product development process from the outset.

Google Earth will almost certainly remain an invaluable public resource. But by merging trusted geographic reference imagery with on-demand AI generation, Google risked changing more than a product. It threatened assumptions that underpin visual verification itself. If companies want to build tools that become part of the world's information infrastructure, they must stop treating them as consumer creativity products, and must be governed accordingly as systems that shape public trust, reporting, and accountability.

So what comes now? Google says it will work on better “guardrails.” But are guardrails enough when the fundamental accessibility, ease and promotion of the tool itself undermine our shared trust infrastructure in favor of novelty, humorous overlays, dicey fakes and easier real estate visualization?

New York City is working to answer a critical question facing cities and states: How do we put people first in a modern, digital era to fundamentally change how government operates—and what can people expect from it?

What’s most notable is the way the city is answering this question: by giving a new dedicated digital delivery team the internal support, empowerment, and structural authority to prioritize people's needs over bureaucratic processes.

—nicknamed “the PIT Crew.” The PIT Crew is a new government team made up of product managers, designers, engineers, user researchers and data experts to team up with City agencies to build digital solutions to thorny public problems. One of the Crew’s first projects will be to partner with the Department of Consumer and Worker Protection to build an online service where New Yorkers can easily file complaints against companies that place customers in hard-to-cancel, costly subscriptions.

For New Yorkers, that work is the point of government. It's also a test of whether government can reset what people expect from public services both in New York and around the country.

talking to more than 100 government leaders

 who have delivered for people across local, state, and federal levels. We ourselves have built, led, and supported successful delivery teams around the country, including leading IRS’ Direct File. What we’ve learned is that when teams can organize around outcomes, not bureaucracy, they can truly move at the speed that government’s biggest challenges require.

New York City is looking to do this at scale. What can your city learn from the NYC PIT Crew—not just the team, but the theory of transformation? The answer lies in five principles that should become the standard operating procedure for digital capacity initiatives everywhere.

1. Set a standard that changes everything

Most government digital projects aim for incremental improvements, setting a goal of “better than last time.” That’s insufficient. The IRS Direct File showed what’s possible when you aim higher: services so intuitive, accessible, and reliable that residents come to expect government to just work.

Rather than settling for mere “improvement,” New York City—and other cities and states—should adopt aggressive user experience standards and measure success against private-sector benchmarks, not against other government agencies.

2. Organize around people’s needs, not org charts

Government is siloed by design. People don't experience government by agency. They experience it in moments: applying for a job, starting a business, navigating a health crisis. When a new parent needs childcare support, they don't care whether it's "Housing's job" or "Labor's job." They care that it works.

The most effective delivery teams organize themselves around people’s journeys. They break down silos in funding, staffing, and operations. Every state and city that wants to improve outcomes should ask: 

Are we organized to solve problems for residents, or for ourselves?

3. Empower teams to own outcomes end-to-end

Accountability in government is notoriously diffuse. This is where most initiatives die—not on launch day, but six months later, when the process starves what shipped. Successful digital capacity initiatives change that. A single team owns a service, is transparent about performance, and has the authority to make real changes. This requires giving teams:

Clear responsibility for measurable outcomes

Authority to hire, procure, and design without permission layers

Budget control to avoid the classic government trap of funding being held hostage by competing departments

Local governments serious about scaling impact should pilot this model in one high-impact service and measure results after 90 days.

This is where most digital initiatives fail: they build something beautiful and then get strangled by traditional government processes. Budget cycles, hiring freezes, and procurement bureaucracy can kill momentum faster than any technical failure.

The answer is straightforward in theory, hard in practice: apply agile, outcomes-driven approaches to 

—hiring, procurement, budgeting and policy approval. New York City's commitment to this approach through its 

Commission on Government Efficiency (COGE)

 is encouraging. Other governments should move faster in this direction.

5. Build openly and aim for replicability

When government solves a problem well, it should share the solution. Open-source code, reusable service components, documented playbooks—one city's solution becomes every city's starting point. One city shouldn't have to reinvent the wheel when another has already built it.

Governments should compete on outcomes, not on proprietary approaches. The team that figures out how to streamline unemployment benefits should publish that playbook. The agency that builds accessible design systems should share them.

The launch of the New York City PIT Crew is encouraging because it signals that leadership understands what’s needed: in-house talent, organizational authority, and political cover. That’s rare. Most governments talk about “innovation” while structurally making it impossible.

But the launch of a team isn't success—delivery is. And delivery at scale requires the principles above.

The opportunity in front of governors, mayors, and city administrators is clear: Don't wait to see if New York City's model works. Begin building your own version now, before a disaster strikes and you need your own PIT crew to clean it up. Organize a team around the highest-impact resident experience problem in your area. Give them authority. Measure relentlessly. Share what you learn.

People in other states and cities shouldn't have to wait for their government to deliver what New Yorkers will soon expect as standard. That gap—between what's possible and what's normal—is where innovation lives.

It's now up to other governments around the country to close it.

Telegram on App Store displayed on a phone screen and French flag displayed on a screen in the background are seen this illustration photo taken in Krakow, Poland on December 3, 2023. (Photo by Jakub Porzycki/NurPhoto via AP)

A multilingual network of Telegram channels offering local news does not appear to be what it seems. It seems there is no newsroom behind any of it. Independent researchers working separately, including us at Democracy Reporting International, have reached the same conclusion about this cluster of growing Telegram channels: They belong to a coordinated operation spreading disinformation and propaganda aligned with the Kremlin that shares the same script, pushes biased and unverified content, translated and timed to hit everywhere at once.

Our research reveals how easily a network can significantly grow across many languages in a way that is difficult to detect, which is why we argue that European authorities should reconsider their light regulatory approach to Telegram.

During social media monitoring of the Hungarian elections earlier this year, DRI 

 seven of those previously uncovered channels, along with an additional twelve accounts that nobody had connected to this network. All 19 channels share the same signature of coordinated inauthentic behavior, and they never link or mention one another. Yet they all have similar names (translating roughly to "common sense"), and all but one share identical branding.

The "Common Sense network,” as we call it, shows two kinds of growth: vertical (rising subscriber counts within existing channels) and horizontal (new channels added over time). They tell different stories.

Subscriber numbers function as social proof. At its peak last winter, the network had over 220,000 subscribers; by early April it had dropped to 167,000, and it now sits under 100,000. We are confident the number will rise again, as it did last year, when each channel suddenly surpassed 10,000 subscribers almost overnight. The sudden spikes are consistent with purchased or otherwise inauthentic accounts. However, an unsuspecting German reader who does not spend time verifying a Telegram channel's legitimacy, and who encounters a channel "Die Stimme der Vernunft" which is posting regularly and has 40,000 subscribers, will likely conclude it is a serious channel. There is little to indicate that he or she has just walked into a network repeatedly linked to foreign information manipulation and interference (FIMI) operations.

Horizontal growth is a less ambiguous indicator. Nineteen channels across nineteen languages are themselves a significant footprint, and the more revealing insight is how the network came together. We reconstructed its horizontal expansion using data from the third-party service Telemetrio combined with systematic interval sampling of published messages via the Telegram API.

Step-line chart titled 'Network Growth Timeline (2022–2026): 19 channels across 19 languages,' showing the cumulative number of Telegram channels in the Common Sense network rising from 0 to 19 between November 2021 and May 2026. Growth occurs in discrete jumps: March 2022 adds 2 channels (PL, CS); June 2023 adds 1 (DE); August 2023 adds 6 (ES, HU, FR, IT, EL, NL); July 2024 adds 1 (EN); October 2024 adds 1 (SK); March 2025 adds 1 (RO); July 2025 adds 6 (SR, BG, HE, TR, FI, AR); and March 2026 adds 1 (SW). A shaded green region marks confirmed dates through mid-2025, and a shaded orange region marks partially estimated dates after August 2025. The largest single-day expansions—six new language channels each—occurred in August 2023 and July 2025.

Reconstruction of the Common Sense network's horizontal growth (N=19). The network may include additional channels that went undetected. Creation dates for confirmed channels are known exactly; for estimated channels, dates are approximated to a two-week window.

None of this growth, in either direction, is likely to be noticed by Telegram. A network that can double its subscriber count overnight, or add six new language audiences within days (as it did in August 2023 or July 2025) operates without triggering any visible intervention by the platform.

The obvious question is why such a network can continue expanding despite repeated public exposure.

The story with Telegram and EU regulation is also about numbers. Under the DSA, a platform becomes a Very-Large Online Platform (VLOP) if it has more than 45 million active monthly users. Telegram's 

 states they are still below that threshold. Its obligations are therefore largely limited to tackling illegal content: providing users with mechanisms to report it, acting on those reports, and protecting minors' privacy and safety. Very Large Online Platforms and Search Engines (VLOPSEs) obligations, such as identifying and mitigating systemic risks, do not apply.

The “common sense network” relies on purchased bots, fake engagement, and automated or semi-automated activity to artificially amplify certain content and create the false impression of broader public interest than it actually has. While the content is not illegal, such inauthentic “news channels” undermine the integrity of the public debate. If Telegram were designated a VLOP, it would have to assess and mitigate systemic risks associated with coordinated manipulation.

Nothing, however, prevents Telegram from moderating this type of behavior voluntarily. 

, another non-VLOP, already prohibits harassment, spam, disruptive behavior, impersonation, and other forms of coordinated abuse under its community rules. Telegram has instead chosen a far more hands-off approach, at times even resisting l

Even where Telegram’s obligations are clear, its moderation record falls short of comparable platforms. On July 23, the 

 over the past 180 days. A higher volume of moderation doesn't necessarily mean better protection for users. It can also reflect more user reporting or over-removal. Even so, the gap with its closest counterpart is significant: WhatsApp Channels, comparable in design and user base, recorded 

 over the same period -roughly sixteen times as many.

Nor can this gap be attributed to a lack of content requiring moderation. Civil society organizations have recently documented illegal content on Telegram, including 

 (NCII), and child sexual abuse material (CSAM). Last year, the Belgian Institute for Postal Services and Telecommunications (BIPT) –the Digital Services Coordinator responsible for overseeing Telegram’s compliance with the DSA– separately identified hundreds of 

Taken together, these factors point to a gap: a platform with considerable reach and influence in the European Union is operating under a significantly lighter regulatory burden than its 

A path forward for stronger Telegram accountability

The most direct path is VLOP designation. Pressure is building on the European Commission to examine Telegram's status. In March 2025, a 

 the Commission to clarify the methodology behind Telegram's EU user count, pointing to independent research that suggests the real figure exceeds 45 million. The 

 was noncommittal: it said only that it was monitoring Telegram's figures, following market developments, and remaining in contact with relevant authorities.

Telegram's reported figures do raise legitimate questions. In February 2024, the company 

 average monthly active recipients in the EU. Since then, it has offered only the claim of "less than 45 million," even as its global user base has 

. That the EU figure has stayed just under the threshold throughout this period merits closer examination, particularly since WhatsApp Channels, Telegram's closest equivalent in design and use, crossed the threshold and was 

Part of the problem is methodological. Article 24(2) of the DSA gives the Commission the power to adopt a delegated act detailing the methodology of how "active recipients" should be counted, but it has not yet exercised that power, relying instead on 

. According to this guidance, an “active recipient” is anyone exposed to content disseminated through a platform's online interface. Public Telegram channels can be viewed, searched, indexed, and shared by anyone, whether or not they subscribe. If exposure to content is the relevant standard, as the DSA suggests, there are reasonable grounds to question whether Telegram's reported figures fully capture the reach of its public channels. Greater transparency about how the methodology should be to calculate those figures would help answer that question.

In the meantime, the European Commission and the Belgian regulator have tools available to strengthen Telegram’s accountability. The Belgian regulator could use its supervisory powers to examine Telegram’s compliance with its existing obligations on illegal content and child safety and open enforcement action where necessary. Research from civil society organizations and independent investigators is a crucial source of evidence to push for these efforts.

also shows that content identified on Telegram circulates across VLOPs, including Facebook, X, Instagram, YouTube, and TikTok. While our analysis does not prove coordinated cross-posting, it suggests that disinformation networks operate across multiple platforms. The DSA already provides a systemic risk framework that encourages stronger cooperation between VLOPs, VLOSEs, and platforms such as Telegram to identify and disrupt campaigns that move across services.

Commission's DSA Guidelines on mitigating systemic risks to electoral processes

 can serve as a useful benchmark for platforms such as Telegram. Although they cannot create new legal obligations for non-VLOPs, they can establish clearer expectations for addressing coordinated manipulation and other threats to electoral integrity.

Can the Pall Mall Process Curb Spyware Abuse?

A group of shareholders of Samsung Electronics Co. hold a rally criticizing a tentative wage agreement struck between the company and its labor union near the home of Samsung Electronics Chairman Lee Jae-yong in Seoul, South Korea, Thursday, May 21, 2026. The letters read "Unjustified transfer strike." (AP Photo/Ahn Young-joon)

Workers in Asia Are Fighting for Protections as AI Threatens Jobs

A police officer checks the list of a polling station during the parliamentary elections in capital Nicosia, Cyprus, Sunday, May 24, 2026. (AP Photo/Petros Karadjias)

EU Political Ad Rules Leave Elections Open to Foreign Manipulation

Tech Policy Press - Technology and Democracy

Tech Policy Press is tracking laws and regulations, along with government investigations and litigation, that will shape the rules and accountability for tech companies.

Latest

Podcast

Recent

Our content delivered to your inbox

Thank you!

Around the globe