UK plans takedown rule for tech firms over abusive images

The UK proposes new rules requiring tech firms to remove abusive images within 48 hours, following concerns over AI-generated inappropriate content.

The regulations, announced by the government, could impose fines of up to 10% of global revenue or lead to service bans in the UK for non-compliance.

Ofcom is also considering measures to automatically detect and remove intimate images shared without consent, similar to existing laws on non-consensual sharing.

The move comes after criticism of social media platforms, including Elon Musk’s X, over the distribution of sexualized images generated by AI tools, which prompted the platform to restrict certain features.

Sharing non-consensual intimate images is already illegal in the UK, and the proposed rules aim to strengthen enforcement through amendments to the crime and policing bill.

The UK Safer Internet Centre highlighted challenges in prompt content removal, emphasizing the need for faster action to support victims.

🔗 Source: Bloomberg

🧠 Food for thought

Implications, context, and why it matters.

New UK rules build on the Online Safety Act, but will be introduced via the crime and policing bill

The proposal would sit alongside the UK’s wide-ranging Online Safety Act (OSA), which became law in 2023 1.
The OSA expanded criminal law to cover intimate image abuse, including altered or created images such as deepfakes, based on the Law Commission’s review and recommendations 2.
Ofcom is urging platforms to use automated “hash-matching” technology (a way to identify known images by their unique digital fingerprint) to detect and remove intimate images shared without consent. It is also consulting on making the measure mandatory 3.
The OSA lets Ofcom fine companies up to 10% of qualifying worldwide revenue. In-scope platforms had to complete illegal content risk assessments by 16 March 1.

Global regulators are moving toward shared accountability for high-risk generative AI

The dispute over X’s AI tool has prompted regulators to open probes in several places, including the EU, Canada, and India 4.
Regulators may speed up use of technical standards for AI-generated content. These include digital watermarking (embedding a detectable marker in media) and provenance signals (metadata that helps track where content came from), especially in election security and child safety 4.
The policy debate is widening from harmful posts to the design and operation of AI systems.
Over time, legal frameworks could split accountability across developers, deployers of high-risk generative AI tools, and the social networks where the content appears 4.