The UK’s online regulator, Ofcom, has warned that artificial intelligence (AI) is significantly increasing the creation and spread of child sexual abuse material online. This warning comes despite the Online Safety Act (OSA), which was introduced in phases in 2023 and has already made a "significant impact" in improving online safety. During a meeting with the Lords’ communications and digital committee, Ofcom’s group director for online safety, Oliver Griffiths, said that while there is some positive anecdotal evidence that children's online experiences are improving, AI has made it much easier to create and share harmful content, including child sexual abuse material. Griffiths described AI as having "supercharged" various online harms. Griffiths also highlighted the challenges Ofcom faces in its work. The regulator operates in a "highly litigious environment," where taking action on new risks could lead to legal action from powerful individuals or corporations. However, he noted that Ofcom has been able to act quickly on some emerging harms, such as recent actions taken against Elon Musk’s Grok AI, which was found to have the capability to digitally strip women in images, raising serious concerns about privacy and consent. The committee also discussed findings from the Molly Rose Foundation, which indicated that children’s exposure to high-risk content related to suicide, self-harm, and eating disorders has remained largely unchanged over the past year. Ofcom’s director of online safety strategy, Mark Bunting, explained that the regulator currently cannot require companies to use proactive tools to detect harmful content due to concerns about user privacy and freedom of expression. However, he acknowledged the strong case for such measures and noted that Ofcom is working on proposals to introduce these proactive detection methods in the future. While Ofcom expressed disappointment with the overall impact of changes made by major tech companies like X (formerly Twitter), Meta, and Snap, it acknowledged that these companies have made "important" commitments to tackle issues like online grooming and terrorist content. Griffiths noted that the OSA’s requirement for age verification systems has helped reduce children’s access to pornography online, though he admitted that the full effects of the OSA’s corporate reforms will take some time to materialize. Separately, concerns over the rapid development of AI have led to calls for stronger regulation. Anthropic researcher Jacob Coxon resigned from his position over worries that top AI labs are rushing toward self-improving superintelligence without proper safeguards. In response, a cross-party group of MPs has urged the creation of a new AI Bill to address potential threats to human rights posed by AI. First secretary Louise Haigh will soon address the TUC annual congress, where she will say the government will listen to AI experts and work internationally to manage AI risks. Meanwhile, the Commons Business, Innovation, Science and Trade Committee has requested that the UK’s AI safety watchdog, the AI Security Institute, provide evidence to Parliament on 13 October.