Google plans partnership to fight non-consensual images
Sign up now: Get ST's newsletters delivered to your inbox
Google has been criticised for dragging its feet in adopting a technology that protects victims of image-based abuses.
PHOTO: REUTERS
LONDON – Alphabet’s Google plans to start working with StopNCII, a non-profit that helps prevent the spread of non-consensual images online. For advocates, the move marks a significant, albeit overdue, step in the search giant’s efforts against image-based abuse.
StopNCII’s technology allows victims of image-based abuse to create digital fingerprints, or hashes, of intimate images. These hashes are then shared with partner platforms, including Facebook, Instagram, Reddit and OnlyFans, which use them to prevent reupload of the images without requiring anybody to look at or report them.
Google announced the partnership at the NCII summit on Sept 17 at its London office, where they were hosting StopNCII’s parent charity, SWGfL.
For victims, “knowing that their content doesn’t appear in search, I can’t begin to articulate the impact that this has on individuals”, said Mr David Wright, SWGfL’s chief executive officer, in an interview.
Google will not immediately appear on StopNCII’s official partner list. A Google spokesman said the company is currently testing the technology and expects to begin using the hashes “over the next few months”.
Adopting the hash matching technology would be a significant change that would require evolving the company’s processes and infrastructure over time.
Google’s comparatively slow pace in adapting the technology has drawn criticism. StopNCII launched in late 2021, building on detection tools used at Meta.
Facebook and Instagram were among the early partners, TikTok and Bumble joined in December 2022, while Microsoft integrated the system into its Bing search engine in September 2024, nearly a year ahead of Google.
When questioned explicitly on why the company was not yet adapting StopNCII’s technology, Google told British lawmakers in April 2024 it had “policy and practical concerns about the interoperability of the database” and, therefore, could not yet participate.
Not far enough
Some advocates say Google’s move does not go far enough.
“It’s a step in the right direction,” said Mr Adam Dodge, founder of advocacy group End Technology-Enabled Abuse. “But I think this still puts a burden on victims to self-report.”
A company with Google’s resources, he argued, could go further in taking non-consensual images down, without requiring victims to create hashes.
Missing from Google’s announcement is any mention of AI-based non-consensual imagery, or deepfakes. StopNCII relies on known images, meaning it cannot preemptively block AI-generated deepfakes.
“If it’s a synthetic or an entirely different image, the hash is not going to trap it,” Mr Wright said.
In 2023, Bloomberg found that Google Search was the top traffic-driver to websites hosting deepfakes or sexually explicit AI-generated pornography. Since then, the company has taken some steps to reduce and downrank the content in search. BLOOMBERG