SoundCloud Product Lead Confronts Platform's AI Bot Problem: "Continuous Game of Cat and Mouse"

A SoundCloud product lead has offered the company’s most direct public accounting yet of its long-running bot problem, conceding that artificial intelligence has made automated accounts harder to catch.The remarks came from SoundCloud’s Principal Product Manager, Harry Manion, who is set to host an AMA (“ask me anything”) in the company’s verified subreddit.

The session is scheduled to run on September 14th, but Manion dropped by early to answer a few questions on Friday.When asked about bots, he said he has spent considerable time on the issue, including going undercover in Telegram groups run by SoundCloud bot makers.

He framed the stakes plainly, writing that “every 1 bot experience devalues the next 10 real ones you get.” By Manion’s account, SoundCloud now detects and blocks “millions” of inauthentic actions per week, a figure he offered as evidence of progress over the last couple of years.But he acknowledged that AI has helped bots defeat previously secure systems and blur the line between real users and automated ones, a “continuous game of cat and mouse” the platform has to actively monitor.

Manion said he’s hesitant to share specifics on the company’s plans, instead pointing to three broad areas of work to be done: making bot accounts harder to create and maintain, building better trust signals out of user behavior, and giving users easier ways to filter and flag the spam that gets through.“On the flip side,” he continued, “it’s equally important that *real* SC’ers understand that simple things like having a profile picture, a bio, dropping a comment, and messaging a random person to say you like their track or playlist, all make the platform feel more human and connected.

Plenty of people exist as empty profiles, which means that genuine actions can feel spammy to the person receiving them.” You can check out the full AMA here starting September 14th at 2pm ET (11am PT).Scroll to continue ...