"This lack of connection between what we commit to publicly and what we do internally really worried, and slightly shocked, me."

Molly Cutler: "On January 22, the BBC reported that exposure to graphic images of cutting and self-injury on Instagram contributed to the suicide of 14-year-old Molly Russell. Our policy permitted these images unless they clearly promoted or celebrated suicide or self-injury. Graphic images shared with no caption or context did not violate our policy. These non-violating images were discoverable in Instagram Explore and did not count in the >30% threshold used to implement the 'hashtag playbook' on Instagram, which reduces the discoverability of hashtags with a high proportion of violating content. We faced several weeks of intense press and policymaker criticism, largely confined to the UK, focused primarily on the availability of this content on Instagram, and secondarily on the discoverability of this content on Instagram. Throughout this period, our product policy team appeared confident that our policy permitting graphic cutting content was likely to remain largely in place, as informed by consistent engagement with experts on these issues, and we worked to pursue a long-term product strategy of reducing the discoverability of non-violating graphic cutting content, an incomplete but meaningful response to the criticism. Starting on January 26, in response to escalating criticism, we said that we would 'undertake a full review of our enforcement policies and technologies' around self-harm. On February 4, we met with experts and decided to change our policy. Going forward, we will remove graphic cutting images even in the context of admitting one's mental health struggles, because we now recognize that the potential for modeling how to self-harm, normalizing self-harm, and/or triggering self-harm in others outweighs the value of sharing the content. In addition, we will make product changes to reduce the discoverability of non-graphic self-harm content—such as healed scars—by preventing it from appearing in search, hashtags, the explore tab, or recommendations. These announced changes have landed well and we got credit—but do not undo the damage we have incurred as a result of this protracted cycle." "We had the wrong policy on graphic cutting images, and were too slow to reconsider/schedule engagement with experts about our policy in response to escalating criticism." "We strategically de-prioritized SSI product work on IG. This was due in part to the previous narrow policy resulting in low impact of work—relatively small proportion of content eligible for removal under previous policy." Nick Clegg: "Thx Molly – a thankless task, but am grateful to you for doing this. It's all accurate – except that four days after the BBC story broke I said publicly (carried in all UK headlines on the day) that we would review all our policies, and that we would do 'whatever it takes' to address the legitimate concerns raised. Javi, Guy and I were in touch by email – and I had alerted Sheryl, Mark and even the Board to the gravity of the issue. Yet, despite Guy's team's great cooperation, it was not till the large meeting in Sheryl's office – I can't remember the date – with myself, Chris C, Schrep, Dave Wehner and others that there was any recognition that the policy was too heavily skewed towards the needs of those committing SSI rather than those susceptible to copying it. This lack of connection between what we commit to publicly and what we do internally really worried, and slightly shocked, me. I don't know how/whether you want to reflect any of this – but it is important context."