X Claims Record Engagement, Leaks Brand Safety Pitch
X (formerly Twitter) is touting all-time high user engagement, hitting records twice in two days. Meanwhile, a leaked deck reveals the company is using its "Grok" AI to push a new brand-safety narrative to win back advertisers.
The new brand safety pitch comes as X claims it has hit its highest usage of all time, with Elon Musk and Head of Product Nikita Bier stating the platform broke records on February 28th and March 1st, 2026. The surge in activity, attributed to the US-Iran conflict, reportedly peaked at 417 billion global "user-seconds" on a single day. This "user-seconds" metric is a relatively new and not independently verified measure of engagement for the platform. While X is touting these record numbers, CEO Linda Yaccarino acknowledged in late 2023 that daily active users had declined to 225 million, down from 254.5 million before Elon Musk's acquisition. Some external analyses in early 2026 place the number of daily active users even lower, at around 141.5 million. The push to win back advertisers centers on using Grok AI to score content for brand safety, with X claiming it can achieve nearly 100% brand-safe scores as measured by independent auditors like IAS and DoubleVerify. This is a significant shift for a company that previously took legal action against advertisers for using blocklists to prevent their ads from appearing next to certain content. This charm offensive is aimed at reversing a massive decline in advertiser confidence and revenue. Since 2022, marketers' trust in X as a safe platform for ads has plummeted, with one study finding only 4% of marketers believed X provides brand safety. This exodus led to advertising revenue collapsing from a peak of over $5 billion in 2021 to an estimated $2.5 billion in 2024. The leaked presentation was shared with clients and agencies during a "2026 Brand Suitability Webinar" on February 26th. It highlights Grok's ability to review user profiles and posts for sensitive topics, and allows advertisers to block up to 4,000 keywords and 2,000 specific author handles from their campaigns.