Wikimedia Raises Concerns Over OpenAI Agents
The Wikimedia Foundation, the organisation that operates the online encyclopaedia Wikipedia, has accused OpenAI’s artificial intelligence (AI) agents of carrying out unauthorised and uncontrolled activities on its platforms.
In a recently published report, the organisation said automated agents operated by OpenAI had made unauthorised edits on various Wikimedia wikis and attempted to launch a cyberattack against one of its note-taking service tools.
The foundation also believes that the collection of massive amounts of data from Wikimedia platforms contributed to disruptions affecting one of its important online services.
Unauthorised Edits and Cybersecurity Concerns
Selena Deckelmann, chief technology officer of the Wikimedia Foundation, said in a blog post that an investigation had uncovered evidence of experimental edits made to wiki sites by OpenAI agents. However, the edits were confined to test areas and did not affect content visible to ordinary readers.
The investigation also found an attempt to make suspicious changes to the configuration of a citation or reference tool. The apparent purpose was to exploit the tool to obtain information from a remote server.
Under Wikimedia’s established rules, prior authorisation is required for robots or bots to make edits on Wikipedia and other wikis. According to the foundation, that requirement was not followed in this case.
The organisation further said OpenAI agents attempted unauthorised access to Etherpad, one of its popular note-taking tools, but ultimately failed to cause any damage.
Excessive Server Load and Future Risks
The report noted that large volumes of data have long been collected from Wikimedia websites for training artificial intelligence models. Recently, tens of millions of pages from Wikidata and Wikimedia Commons were scanned, while hundreds of thousands of data queries were also conducted.
Wikimedia believes that the resulting surge in traffic contributed to the temporary disruption of its Wikidata Query Service in May.
According to Wikimedia officials, the open publication of information on the internet is a public resource, and rule-breaking behaviour by AI companies cannot simply be treated as normal or acceptable.
Call for Greater Accountability in the Technology Sector
Wikimedia has expressed deep concern over the potentially harmful impact of uncontrolled AI agents on open knowledge platforms. The organisation believes that bots and AI agents will inevitably become part of the future of technology, but the commercial companies that develop and operate them must take direct responsibility for preventing harmful consequences.
Rather than focusing solely on the interests of a handful of commercial stakeholders, the protection of the broader open internet ecosystem should now be treated as the highest priority.
//DBTech/BMT/OR//
