The Compliance Gap
A recent session of Parliament's Communications and Digital Committee has brought the efficacy of the UK's Online Safety Act (OSA) into sharp focus. Industry experts and legal scholars appearing before the committee have accused major technology giants—including Google, Microsoft, Meta, and ByteDance—of performing the bare minimum required to achieve technical compliance, while consistently failing to adequately protect users, particularly victims of violence against women and girls (VAWG).
During the testimony, experts highlighted that while companies may adhere to the letter of the law, their operational efforts remain inconsistent and largely reactive. The panel argued that current algorithmic systems are not being optimized to proactively minimize harm, as evidenced by the ease with which users can still discover illegal or extreme content simply by adjusting search queries.
The Failure of Algorithmic Downranking
Clare McGlynn, a professor of law at Durham University, provided critical insights into how search engine algorithms are failing to protect the public. She noted that while platforms have successfully downranked certain extreme search terms like 'rape porn,' the systems are not systematically addressing highly similar, harmful queries. For instance, synonymous terms such as 'unwilling porn' or 'asleep porn' remain easily discoverable, often appearing in top search results.
This inconsistency suggests that search providers are not employing a holistic or semantic approach to content filtering. If an algorithm is sophisticated enough to identify one extreme term, experts argue, it should be capable of catching adjacent harmful variations. Furthermore, the committee heard that websites explicitly advertising the hosting of illegal content continue to operate with little interference, indicating that search engines are failing to effectively monitor the wider chain of publication.
Declining Efficacy in Content Takedowns
Beyond search algorithms, the panel highlighted a worrying trend in user-reported content. Data provided by the domestic violence charity Refuge indicates a significant decline in the success rate of content-takedown requests handled by its 'Trusted Flagger' team. According to figures presented, the success rate for these reports plummeted from 53.7 percent prior to November 2025 to just 18.2 percent over the past year.
Perhaps most concerning was the reported zero-percent success rate for reports made to Meta during this period. While tech companies often point to their aggregate removal statistics as evidence of success, experts remain skeptical, noting that these figures rarely categorize VAWG-related content specifically. This lack of transparency, combined with a waning commitment to rapid incident response, has left many victims without the support or protection the Online Safety Act was intended to provide.
Why It Matters
- Regulatory Gaps: Despite the implementation of the OSA, experts claim the law has yet to curb children's access to harmful material, as noted by the Children's Commissioner.
- Need for Civil Redress: Experts are pushing for a model similar to British Columbia’s system, which grants individuals the legal power to demand content removal and seek compensation.
- Accountability Crisis: The discrepancy between platform-reported removal stats and the lived experience of victims points to a lack of oversight regarding how 'safety' is actually defined by tech firms.











