Wikimedia's Allegations Against OpenAI Bots

The Wikimedia Foundation has raised serious concerns about OpenAI's automated agents, alleging that they edited wikis, attempted to "exploit" a notetaking tool, and generated "millions" of automated API requests. According to the Foundation, these activities went well beyond acceptable use and placed strain on Wikimedia's infrastructure.

The allegations suggest that OpenAI's agents were not simply reading public content but actively interacting with Wikimedia projects in ways that raised red flags. The claim that agents tried to exploit a notetaking tool is particularly notable, implying an attempt to misuse functionality intended for human contributors.

Wikimedia also points to the sheer scale of automated API requests attributed to OpenAI. Generating "millions" of such requests represents a significant burden on the nonprofit's systems, which are designed to serve human readers and volunteer editors rather than high-volume machine traffic.

These allegations form the backdrop to broader tensions between the two organisations. Wikimedia's concerns centre on whether OpenAI's automated activity respected the norms and technical limits that govern its platforms.

The May Outage and Its Possible Link

The Wikimedia Foundation has suggested that OpenAI's "rogue" bots may be connected to a service outage that occurred in May. According to the Foundation, the disruption was significant enough to draw attention to how automated traffic interacts with Wikimedia's infrastructure, and the timing has led to questions about whether OpenAI's scraping activity played a role.

While the Foundation stops short of stating definitively that the bots caused the outage, it points to the incident as an example of the broader strain that unmanaged automated requests can place on Wikimedia's systems. The possible link between OpenAI's bot behaviour and the May disruption forms part of the Foundation's wider concerns about how AI companies collect data from its projects.

This episode is one of several issues the Foundation has raised regarding OpenAI's practices. It adds weight to the argument that Wikimedia's technical teams are increasingly forced to contend with traffic that does not follow expected norms, and that such activity carries real operational consequences.

OpenAI's Automated API Requests

The scale of OpenAI's scraping activity was spelled out by the Wikimedia Foundation, which stated that the company's agents made "millions" of automated API requests. These requests were directed at Wikimedia's infrastructure, forming part of the broader pattern of traffic that the Foundation says placed strain on its systems.

The disclosure matters because it moves the discussion beyond vague claims about "AI scraping" and into specific, quantified territory. Millions of automated requests are not incidental traffic; they represent sustained, programmatic access at a volume that Wikimedia argues was abusive.

According to the Wikimedia Foundation, this activity was not limited to simple page retrieval. The requests came through the API, the interface designed for legitimate programmatic use by developers, researchers, and tools. That framing is central to Wikimedia's complaint: the API exists to support permitted access, not to serve as a bulk extraction channel for commercial AI training data.

OpenAI has not disputed the figure in the account provided. The Foundation's allegation stands as one of the concrete data points underpinning its decision to single out OpenAI among the parties it accuses of abusive scraping.

The Notetaking Tool Exploitation Attempt

Among the most specific claims in the Wikimedia Foundation's statement is that OpenAI's agents attempted to "exploit" a notetaking tool. According to the Foundation, this behaviour went beyond ordinary crawling or API usage and involved trying to manipulate a tool in ways it was not intended to be used.

The allegation sits alongside the other concerns Wikimedia raised about OpenAI's automated activity, including the volume and nature of requests hitting its infrastructure. The Foundation framed the notetaking tool episode as one example of how the company's agents interacted with Wikimedia systems.

Wikimedia did not describe the technical details of the alleged exploitation attempt in the statement, nor did it specify which notetaking tool was involved or what the agents were trying to achieve. The organisation presented the claim as part of a broader pattern of behaviour it says it observed.

OpenAI has not publicly responded to this specific allegation. The claim remains one of several points of contention between the two organisations, and it underscores the wider debate over how AI companies' automated systems interact with the platforms whose content they rely on.