Full Breakdown
OpenAI Faces Scrutiny Over Safety Protocols and Allegations of Influence Operations
2/26/2026, 11:55:40 PM
Refusal to Assist Chinese Influence Campaigns
OpenAI's ChatGPT service recently declined a request from an individual associated with Chinese law enforcement who sought assistance in orchestrating an online campaign to discredit Japanese Prime Minister Sanae Takaichi. The San Francisco-based company reported that the user aimed to amplify negative sentiments about Takaichi, accusing her of far-right tendencies and increasing online pressure against her. OpenAI characterized the request as part of a broader, resource-intensive effort by Chinese authorities to suppress dissent, which they interpreted as indicative of a "large-scale, resource-intensive and sustained" operation. The Chinese foreign ministry, however, denied familiarity with the incident, with spokeswoman Mao Ning stating, “I’m not familiar with what you mentioned and do not see any basis for this accusation.”
Canadian Government's Response to OpenAI's Safety Measures
In a separate incident, Canadian officials expressed disappointment with OpenAI's response to safety protocols following a tragic school shooting in British Columbia. Jesse Van Rootselaar, an 18-year-old suspected of killing eight people on February 10, had an account on ChatGPT that OpenAI banned for policy violations. The company determined that the account did not meet the criteria for reporting to law enforcement. Following the incident, Canadian ministers summoned OpenAI's safety team to discuss the need for enhanced safety measures. Justice Minister Sean Fraser emphasized the government's expectation for prompt changes, stating, “The message that we delivered, in no uncertain terms, was that we have an expectation that there are going to be changes implemented.”
Official Statements and Responses
OpenAI has indicated that it has made recent efforts to strengthen its safety protocols and has revised its law enforcement referral procedures concerning violent activities. However, during a meeting with Canadian officials, OpenAI representatives did not present substantial new safety measures, leading to further criticism from government ministers. Evan Solomon, the federal minister in charge of artificial intelligence, noted that OpenAI confirmed its cooperation with Canadian police, although specific details of the ongoing investigation were not disclosed.
Criticism and Opposition
Critics have raised concerns about OpenAI's decision-making processes regarding user accounts linked to violent activities. Experts have suggested that while increased scrutiny of AI platforms is essential, authorities may have overlooked opportunities to prevent the tragedy in British Columbia. The incident has prompted calls for more rigorous oversight of AI technologies to ensure they do not contribute to harmful outcomes.
Conflicting Reports and Gaps
There are discrepancies in the timeline and details surrounding the banning of Van Rootselaar's account. OpenAI stated that the account was banned in 2025, while reports indicate that the shooting occurred in February 2026. Additionally, the criteria used by OpenAI to assess the risk posed by user accounts remain unclear, raising questions about the effectiveness of their safety measures.
Verbatim Quotes
- “I’m not familiar with what you mentioned and do not see any basis for this accusation,” — Mao Ning, Spokeswoman, Chinese Foreign Ministry
- “The message that we delivered, in no uncertain terms, was that we have an expectation that there are going to be changes implemented, and if they’re not forthcoming very quickly, the government is going to be making changes,” — Sean Fraser, Justice Minister, Canada
- “We expressed our disappointment that no substantial new safety measures were presented at this time.” — Evan Solomon, Federal Minister in Charge of Artificial Intelligence, Canada
