A shared fraud intelligence database co-founded by Google today released its most significant platform update and new research confirming the problem it was built to solve is larger than anyone had measured. For marketing professionals, the implications run closer to home than they might expect.

What the Global Signal Exchange actually is

Start here, because nothing else in this article makes sense without it.

When a scammer runs a fake investment ad, sends a phishing text, or registers a fraudulent domain, that activity leaves traces. A bank sees a spike in customers reporting fake transfer requests. A telecoms company sees unusual SMS patterns. A search engine sees a cluster of newly registered domains mimicking legitimate brands. An ad platform sees a campaign that violates its policies. Each organisation collects those traces - call them threat signals - and uses them to protect its own users. But almost none of them share those signals with each other, even when the scammer is targeting all of their users simultaneously, using the same infrastructure.

The Global Signal Exchange (GSE) was built to fix that. Launched in January 2025, it is a UK-based non-profit owned by Oxford Information Labs (OXIL) - a technology research organisation - with Google as its co-founding partner. The concept is a shared clearing house: member organisations contribute their threat signals to a common pool, and in return gain visibility into signals from every other member. A bank that spots a fraudulent domain can flag it to the GSE. An ad platform can then cross-reference that flag before serving inventory on the same domain. A telecoms provider can use it to block associated phone numbers. The intelligence multiplies because it is shared rather than siloed.

The GSE launched in October 2024 as a pilot with Google contributing data on fraudulent Shopping merchants, and the DNS Research Federation contributing its existing database of around 40 million domain-based signals. By September 2025, Singapore became the first government to join, at which point the platform was tracking 380 million threat signals in real time. Today, that number stands at 1.2 billion.

The problem the GSE was built to solve is bigger than first measured

Today, a new independent study makes the case for why the GSE matters - and why the gap it is trying to close remains largely open.

The Future of Financial Intelligence Sharing (FFIS) research programme, part of the Royal United Services Institute (RUSI) - a British defence and security think tank - surveyed 37 collaborative analytics platforms tackling fraud worldwide. The study was released by researcher Nick Maxwell. Its headline finding is not that fraud is growing, though it is. The headline finding is structural: 91% of anti-fraud leaders consider cross-border data sharing essential, and yet most platforms that exist to fight fraud remain strictly domestic in their operation.

That gap matters because fraud is not domestic. A scammer operating from one country targets victims in another, routes money through a third, and uses advertising infrastructure headquartered in a fourth. Each organisation defending against part of that chain holds a piece of the picture. None holds the whole thing. According to the FFIS study, fraud cost the global economy an estimated $579.4 billion in 2025. The GSE is identified in the report as a collaborative technical template for bridging that gap - one of very few platforms attempting to operate at the geographic and cross-sector scale the problem actually requires.

Emily Taylor, CEO of OXIL, described the barrier plainly: "The FFIS study confirms what we see every day: the will to share intelligence across borders is there, but the infrastructure to do it hasn't kept pace. Data sharing is easy to commit to in principle, but the legal and technical barriers are real. Jurisdictional complexity and data localisation requirements mean that good intentions don't automatically translate into shared intelligence, and that's precisely the gap the GSE is built to close."

The legal and technical barriers Taylor names are worth understanding concretely. Data localisation requirements in various jurisdictions compel organisations to store and process data within specific national boundaries. A European bank's fraud signals may not legally leave the EU without meeting the conditions of GDPR's Chapter V transfer rules. A government agency's intelligence may be classified or subject to national security constraints. Even where data can technically be shared, liability questions arise: if a signal contributes to a wrongful takedown, who bears responsibility? These are not hypothetical concerns. They are the reasons most platforms surveyed by FFIS, despite every intention to collaborate, remain domestic.

What this has to do with advertising

For marketing and advertising professionals, fraud tends to register as a campaign-level problem: click fraud, invalid traffic, fake installs. The GSE operates at a different layer - the infrastructure layer - but the connection to advertising is direct.

The FTC reported in April 2026 that Americans lost $2.1 billion to social media scams in 2025, an eightfold increase from $261 million in 2020. A substantial portion of those losses flowed through paid advertising. Scam operators are media buyers. They purchase inventory through the same auctions that legitimate advertisers use. They outbid legitimate campaigns for placements adjacent to content that attracts financially distressed users. They register domains that pass automated brand safety checks, serve ads through those domains, and collect personal or financial data from the people who click.

When a programmatic buyer purchases inventory without visibility into the domain-level fraud signals the GSE aggregates, part of their budget may be funding the same infrastructure that defrauds their customers. That is not an abstract concern. A technical analysis published in April 2026 demonstrated that Google Ads IP exclusion lists - one of the primary defensive tools available to advertisers trying to block fraudulent traffic - can be bypassed through a two-step method that exploits the gap between the IP that generates a click ID and the IP that consumes it. The fraud is invisible to the advertiser because the click identifier itself is legitimate.

The GSE's domain-based signal database is directly relevant here. The platform's league tables rank fraudulent domains by their contribution to scam supply chains. For ad operations professionals building publisher exclusion lists or running supply-path analysis, that ranking is actionable data - a prioritised view of where to focus blocking efforts rather than a flat list of flagged domains.

What the new research found inside 30 million signals

OXIL Research, the research arm of Oxford Information Labs, published new findings today drawing on nearly 30 million domain-based signals processed through the GSE. The findings have direct implications for anyone running campaigns in regulated categories.

Across all scam material, broad general population content accounts for 34% of total weighted relevance. That figure describes the widest part of the funnel - scam domains designed to attract the largest possible audience, regardless of specific vulnerability. But when the analysis narrows to the most explicitly harmful signals, the picture changes sharply. Gambling content accounts for 48.1% of that concentrated subset.

More specifically, the research identified a significant tactical overlap between the methods used to target adults with gambling addictions and those deployed against adolescents in the same digital spaces. The same domain infrastructure, the same traffic-driving techniques, and the same persuasion mechanisms appear across both groups. For advertisers operating in gambling, financial services, or any category adjacent to vulnerable audiences, this finding has compliance implications that extend beyond brand safety into regulatory territory. Regulators in multiple jurisdictions are actively examining the degree to which platforms and advertisers bear responsibility for the audiences their inventory reaches.

A second, forthcoming OXIL consumer research study adds a further layer. Preliminary data suggests that people with cognitive or memory disabilities were more than twice as likely to lose money to scams compared to the national average. And 71% of those who experienced a scam did not attempt to report it. That non-reporting figure is particularly important for anyone relying on platform-reported fraud data to understand the problem. If nearly three quarters of victims never contact an authority, every dataset built on reports - including the invalid traffic measurements that verification vendors provide to advertisers - is a significant undercount of what is actually happening.

A March 2026 OXIL study drawing on 28.6 million signals already overturned one assumption: that older adults are the primary targets of online fraud. Working-age adults ranked first among targeted groups. Older adults ranked eleventh. The policy and advertising implications of that finding - which audiences are genuinely at risk, where protective messaging should be concentrated, which placements carry the highest exposure to scam-adjacent content - remain underdigested across the industry.

What the platform update includes and why it matters

Alongside the research announcements, the GSE today released platform version 2.7.0. Three changes are worth understanding in detail.

The first is a Feedback API for at-scale data ingestion. An API-based feedback mechanism might sound like an internal engineering improvement. It is more significant than that. The GSE's value depends on a feedback loop: contributing organisations need to be able to report back on whether the signals they received actually corresponded to genuine fraud. Without that feedback, the system accumulates volume without improving accuracy. A domain flagged as fraudulent may have been cleaned up; a new fraudulent domain may share characteristics with legitimate ones. The Feedback API allows that outcome data to flow back into the system automatically, at the volume and speed that large organisations generate it, rather than through manual submissions.

The second change is Feedback Metrics dashboards, which give contributing partners visibility into the performance of their own signal contributions. This is an accountability mechanism. Organisations can see whether what they share is proving useful to the network. The network can identify which data sources generate the most actionable intelligence. Over time, that visibility creates an incentive structure that rewards high-quality contributions.

The third change is a set of Compass upgrades. Compass is the GSE's investigation interface - the tool partners use to query the signal database and build intelligence products. The 2.7.0 update adds league table support directly into Compass, alongside integration with the GSE Index. The GSE Index is an algorithm that ranks entities - domains, infrastructure providers, hosting companies - by their contribution to enabling fraud. Embedding that ranking directly into the investigation interface means an analyst who identifies a suspicious domain can immediately see where it sits in the supply chain of fraud enablement, without moving between tools. The GSE Index algorithm itself has also been updated to better reflect the balance of factors used in its calculations.

Lucien Taylor, Co-Founder and CTO of the GSE, described the intent behind the update: "Reaching 1.2 billion threat signals is meaningful, but only if partners can act on them. Every improvement we make to Compass and the feedback tooling is about reducing the distance between data and action."

The summit appearances

The research and platform update arrive alongside two significant conference appearances. On 13 May 2026, the Google Safety Engineering Center (GSEC) in Zurich hosted the second EMEA Anti-Scams and Fraud Summit. Karen Courington, Google's VP of Trust and Safety, identified the GSE at the summit as "one of the most effective solutions for cross-sector collaboration." Amanda Storey, Google's Managing Director of Trust and Safety, addressed the collaborative argument directly: "Fighting scams requires a united front, and this platform provides exactly that. By sharing threat data cross-sector, we can identify and disrupt scams before they can cause harm."

Lucien Taylor presented at the summit's technical workshop on the Domain Name System and the fraud lifecycle, demonstrating how the GSE league tables use the platform's 1.2 billion signals to map weaknesses within scam supply chains. OXIL Research colleagues presented Rethinking Scam Prevention, the safeguarding study supported by Google.org, during the afternoon sessions.

The GSE leadership team also appeared at the Global Anti Scam Summit in Lisbon, where the focus was on demonstrating impact from the platform's growing partner network. The emphasis on partner stories is relevant context for understanding the signal volume growth. The GSE reached 380 million signals in September 2025 when Singapore joined. Reaching 1.2 billion by June 2026 reflects new partners joining and new data types being contributed - not simply the same partners generating more of the same signals.

Timeline

  • October 9, 2024 - Google, the Global Anti-Scam Alliance, and the DNS Research Federation announce the Global Signal Exchange as a pilot; Google shares 100,000 URLs of fraudulent Shopping merchants; the DNS Research Federation contributes over 40 million domain signals
  • January 2025 - The GSE formally launches as a UK-based non-profit owned by Oxford Information Labs; OXIL Research begins collecting domain-based signals
  • March 4, 2025 - Google launches on-device Gemini Nano AI scam detection in Google Messages and Phone; the company's Fighting Scams in Search Report cites over $1 trillion in global consumer losses in 2024
  • September 3, 2025 - Singapore becomes the first government entity to join the GSE at the Global Anti-Scam Summit Asia 2025; the platform reaches 380 million real-time signals; Southeast Asia records an estimated $23.6 billion in scam losses in the preceding year
  • March 18, 2026 - OXIL Research publishes analysis of 28.6 million domain-based fraud signals, finding working-age adults are the primary targets of online scam infrastructure and older adults rank eleventh
  • April 24, 2026 - A two-IP method bypassing Google Ads IP exclusion lists is documented by Faktica Analytics, exposing a structural gap in campaign-level fraud defences
  • April 27, 2026 - FTC data shows Americans lost $2.1 billion to social media scams in 2025, an eightfold increase from 2020; Facebook accounts for $794 million of the total
  • May 13, 2026 - Second EMEA Anti-Scams and Fraud Summit at the Google Safety Engineering Center in Zurich; Karen Courington and Amanda Storey address the gathering; Lucien Taylor presents on DNS and the fraud lifecycle
  • May 28, 2026 - Meta announces World Cup 2026 scam safety measures using Visa intelligence and the Fraud Intelligence Reciprocal Exchange to dismantle fake gambling sites using FIFA branding
  • June 2026 - GSE leadership presents at the Global Anti Scam Summit in Lisbon, demonstrating partner network impact
  • June 26, 2026 - GSE releases platform version 2.7.0 with Feedback API, Feedback Metrics dashboards, Compass league table support, and updated GSE Index algorithm; FFIS/RUSI study of 37 platforms released finding 91% of anti-fraud leaders back cross-border sharing while most platforms remain domestic; new OXIL findings show gambling content accounts for 48.1% of the most explicitly harmful signals across nearly 30 million domain records

Summary

Who: The Global Signal Exchange (GSE), a UK-based non-profit owned by Oxford Information Labs (OXIL), co-founded with Google. Associated organisations include the Royal United Services Institute (RUSI), which published the FFIS study through researcher Nick Maxwell; Google's Karen Courington (VP of Trust and Safety) and Amanda Storey (Managing Director of Trust and Safety); and Emily Taylor (CEO of OXIL) and Lucien Taylor (Co-Founder and CTO of the GSE).

What: The GSE - a shared clearing house where companies, governments, and platforms pool fraud threat signals - today released platform version 2.7.0 and reached 1.2 billion signals. The update introduces a Feedback API for automated outcome reporting, Feedback Metrics dashboards for partner accountability, and Compass upgrades embedding league table support and GSE Index integration. A simultaneous RUSI study of 37 collaborative fraud platforms found 91% of anti-fraud leaders support cross-border data sharing but most platforms remain domestic. New OXIL research shows gambling content accounts for 48.1% of the most harmful scam signals, and that people with cognitive or memory disabilities are more than twice as likely to lose money to fraud.

When: The platform update, the RUSI study, and the new OXIL research were all announced on 26 June 2026. The EMEA summit at which the GSE was recognised took place on 13 May 2026.

Where: The GSE operates as a UK-based non-profit accessible globally to accredited members. The EMEA summit took place at the Google Safety Engineering Center in Zurich. The Global Anti Scam Summit appearance was in Lisbon. The RUSI study examined platforms across multiple jurisdictions.

Why: Fraud cost the global economy an estimated $579.4 billion in 2025. The organisations best placed to fight it each hold partial intelligence that becomes far more powerful when combined - but jurisdictional complexity, data localisation requirements, and the absence of shared infrastructure keep most of that intelligence locked in domestic silos. For marketing professionals specifically, the same fraudulent domain infrastructure that victimises consumers also sits inside programmatic supply chains, inflating costs and exposing brands to adjacency risk that campaign-level defences alone cannot address.