Profile of Internet Watch Foundation, a UK-based charity helping tech companies identify child sexual abuse material online, as it rebuilds its hashing software
Matt Burgess / Wired UK : Tweets: @wired , @joannashields , @iwfhotline , and @drproudman Tweets: @wired : The Internet Watch Foundation has rebuilt its hashing software, dubbed Intelligrade, to automatically match up images and videos to the rules and laws of Australia, Canada, New Zealand, the US and the UK. https://www.wired.co.uk/... Via @WIREDUK Joanna Shields / @joannashields : Kudos to @IWFhotline for its latest efforts to rebuild its hashing software to automatically match up images & videos to the laws of the #FiveEyes countries - international cooperation & coordination are key in the fight against online child sexual abuse https://www.wired.co.uk/... @iwfhotline : We've rebuilt our hashing software, dubbed Intelligrade, to automatically match up images & videos to the rules & laws of 🇦🇺🇨🇦 🇳🇿🇺🇸 🇬🇧, so it's easier for tech companies to prioritise the most serious imagery of abuse. Story by @mattburgess1 @wireduk. https://www.wired.co.uk/... Dr Charlotte Proudman / @drproudman : Really promising that we're moving towards a more standardised approach of reporting & convicting people for distributing child abuse images. Training on identifying these photos is vital if we want to eliminate paedophilia. https://www.wired.co.uk/...
Context & Ripple Effects
The hash database of one million child sexual abuse images the Internet Watch Foundation built in 2022 has become core detection infrastructure for tech companies — but until now it matched content without regard to where a platform operates. Intelligrade's rebuild changes that: hashes are now automatically graded against the rules and laws of Australia, Canada, New Zealand, the US and the UK, so moderators can prioritize the material that is most serious in each jurisdiction.
The timing matters because the workload is shifting under the IWF: after it first warned about generative AI being used to make abuse imagery, its own counts show the problem compounding — 8,029 realistic AI-generated images and videos identified in 2025, up 14% year over year. A charity-run tooling layer is absorbing work that platforms' own systems have not fully covered, under sustained external pressure like the Heat Initiative campaign pushing Apple on iCloud detection.
First-order effects
- Tech companies using Intelligrade get automatic jurisdiction-specific grading across the five named countries instead of manual legal triage, letting their reviewers act first on the most serious material in each market.
Second-order effects
- Platforms facing activist and regulatory scrutiny over CSAM handling — Apple among them per the related coverage — can point to standardized third-party grading as evidence of diligence, raising the baseline competitors are expected to meet.
- As the IWF's grading becomes the shared reference across FiveEyes jurisdictions, national regulators effectively outsource part of enforcement classification to a single UK charity, concentrating dependency in one nonprofit's methodology.
Third-order effects
- If AI-generated abuse volumes keep climbing along the trajectory the IWF itself reports, hash-matching of known images will be insufficient and the tooling layer will need to grade novel synthetic content — shifting trust-and-safety infrastructure from reactive databases toward predictive classification.
- A pattern of charities supplying cross-border compliance infrastructure points toward de facto private standards bodies sitting between platforms and multiple governments, ahead of any unified statutory framework.
The trend: Cross-border content enforcement is consolidating around nonprofit-run detection infrastructure, with the IWF's hashing tools becoming the shared compliance layer for FiveEyes-country tech platforms.