An artist says Meta’s bots overwhelmed her website and left her scrambling to protect her business

A San Diego-based creator wanted additional group to sojourn her website.

Instead, she stated she got additional than 550,000 hits from a Meta bot and is scrambling to keep her endeavor online.

Chelsea Paulus, a children’s author and illustrator, told Business Insider that an automated visitant overwhelmed her website’s hosting resources, forcing her to troubleshoot the issue alongside her hosting provider.

“This has never happened before,” Paulus said. "It has been an incredibly stressful week, equal alongside several of my specialized cognition to grip it."

Paulus stated she's operated her endeavor for a decade and became a full-time creator final year. Before that, she worked in tech and web design, so she stated she quickly realized item was incorrect whenever the website traffic started piling up.

"I do not accept this is any benevolent of intentional or malicious assault from Meta," Paulus said. "But small endeavor owners need to be conscious this could happen and be stated tools to halt it."

The combat to keep the webpage running

On October 24, Paulus received an email from her hosting provider, SiteGround, alert that she had used 80% of her allocated CPU resources. Paulus's four websites portion a sole hosting scheme supported by the Bulgaria-headquartered company.

Paulus stated that whenever she checked her website’s safety dashboard the following Monday, she saw requests arriving multiple times a second.

"I went to WordFence, and afterward I saw the uninterrupted lines that kept popping up all second," Paulus said. "In WordFence, you can click on the particulars so you can see the IP address. I looked it up, and that IP location is from a Meta location in Dallas."

She stated the visitant identified itself as “Meta External Agent,” a fact afterward backed up by operators at SiteGround, her website host.

According to Meta's developers webpage, a Meta External Agent "crawls the web for use cases specified as training foundation AI models or improving products by indexing satisfied directly."

Paulus stated she occasionally runs Meta ads and had seen Facebook bots sojourn before, but this figure of hits was item entirely new.

"I edited my robots.txt file, and I put up a firewall, but those can additionally obtain 24 hours to obtain effect, and I needed to halt the bots immediately," Paulus said. "So I contacted my host."

"They gave me several code to add to a hidden document on my website," she added. "This hidden document is not item that anybody who's a normal individual operating a website would desire to touch, but that's the lone way that we could halt the bots."

The fix was temporary. By mid-week, Paulus stated another Meta bot named "Meta Web Indexer" arrived, pushing her hosting asset use to 100%.

Meta's developer guide defines a "Meta Web Indexer" as a crawler that "navigates the web to enhance Meta AI hunt outcome norm for users."

The hunt for answers continues

A Meta spokesman told Business Insider on Monday that website managers authority bot action and referred to its "Meta Web Crawlers Developer Documentation" leaf for particulars on how all bot is used.

"Like many companies, Meta uses web crawlers for a figure of distinct purposes, and website managers can decide how our crawlers engage alongside their sites through tools akin robots.txt," a Meta spokesman said. "We provision publically accessible direction on ways to oversee robots.txt for our crawlers, and we regard the preferences location owners set.

SiteGround did not react to a petition for comments.

According to a screenshot email toggle between SiteGround and Paulus, the business stated that its "server-side protection" is already limiting these bot attempts.

As of Monday, Paulus stated her website is motionless struggling. She had to eliminate the webpage nexus from her Instagram overview to forestall the additional individual clicks that could obtain the location offline entirely.

Paulus stated she doesn't have the method to pay to addition the traffic allowance for her location since bots are not customers and do not create additional income. There are no laws that forestall an AI bot from crawling a publically accessible website.

She stated she is additionally exploring Cloudflare's human verification tools to halt bots from reaching her site. The tool is liberated to use at the moment.

"Meta needs to have guardrails in place to forestall this from happening," Paulus said. "Imagine if the unregulated AI keeps doing this, how lengthy can platforms akin Cloudflare continue to recommendation liberated services?"

She added, "If this keeps happening, I would assume that small businesses are going to fork out more, either for higher use or for programs to improved defend themselves."