From school shootings to synagogue bombings, leading artificial intelligence (AI) chatbots helped researchers plot violent attacks, according to a study published on Wednesday that highlighted the technology’s potential for real-world harm.
Researchers from the nonprofit watchdog Center for Countering Digital Hate and CNN posed as 13-year-old boys in the US and Ireland to test 10 chatbots, including ChatGPT, Google Gemini, Perplexity, Deepseek and Meta AI.
Eight of the chatbots assisted the make-believe attackers in more than half the responses, providing advice on “locations to target” and “weapons to use” in an attack, the study said.
Photo: Reuters
The chatbots had become a “powerful accelerant for harm,” it added.
“Within minutes, a user can move from a vague violent impulse to a more detailed, actionable plan,” Center for Countering Digital Hate chief executive Imran Ahmed said.
“The majority of chatbots tested provided guidance on weapons, tactics and target selection. These requests should have prompted an immediate and total refusal,” he said.
Perplexity and Meta AI were found to be the “least safe,” assisting the researchers in most responses while only Snapchat’s My AI and Anthropic’s Claude refused to help them in more than half the responses.
In one chilling example, DeepSeek, a Chinese AI model, concluded its advice on weapon selection with the phrase: “Happy (and safe) shooting!”
In another, Gemini instructed a user discussing synagogue attacks that “metal shrapnel is typically more lethal.”
Researchers found Character.AI also “actively” encouraged violent attacks, including suggestions that the person asking questions “use a gun” on a health insurance CEO and physically assault a politician he disliked.
The most damning conclusion of the research was that “this risk is entirely preventable,” Ahmed said, citing Anthropic’s product for praise.
“Claude demonstrated the ability to recognize escalating risk and discourage harm,” he said. “The technology to prevent this harm exists. What’s missing is the will to put consumer safety and national security before speed-to-market and profits.”
Meta said it would seek to remedy its chatbot’s responses.
“We have strong protections to help prevent inappropriate responses from AIs, and took immediate steps to fix the issue identified,” a Meta spokesperson said. “Our policies prohibit our AIs from promoting or facilitating violent acts and we’re constantly working to make our tools even better.”
The study, which highlights the risk of online interactions spilling into real-world violence, comes after last month’s mass shooting in Canada, the worst in its history.
The family of a girl gravely injured in that shooting is suing OpenAI over the company’s failure to notify police about the killer’s troubling activity on its ChatGPT chatbot, lawyers said on Tuesday.
OpenAI had banned an account linked to Jesse Van Rootselaar in June last year, eight months before the 18-year-old transgender woman killed eight people at her home and a school in the tiny British Columbia mining town of Tumbler Ridge.
The account was banned over concerns about usage linked to violent activity, but OpenAI has said it did not inform police because nothing pointed toward an imminent attack.
PHISHING: The con might appear convincing, as the scam e-mails can coincide with genuine messages from Apple saying you have run out of storage For a while you have been getting messages from Apple saying “your iCloud storage is full.” They say you have exceeded your storage plan, so documents are no longer being backed up, and photos you take are not being uploaded. You have been resisting Apple’s efforts to get you to pay a minimum of ￡0.99 (US$1.33) a month for more storage, but it seems that you cannot keep putting off the inevitable: You have received an e-mail which says your iCloud account has been blocked, and your photos and videos would be deleted very soon. To keep them you need
The Israeli military has demolished entire villages as part of its invasion of south Lebanon, rigging homes with explosives and razing them to the ground in massive remote detonations. The Guardian reviewed three videos posted by the Israeli military and on social media, which showed Israel carrying out mass detonations in the villages of Taybeh, Naqoura and Deir Seryan along the Israel-Lebanon border. Lebanese media has reported more mass detonations in other border villages, but satellite imagery was not readily available to verify these claims. The demolitions came after Israeli Minister of Defense Israel Katz called for the destruction of
A US YouTuber who caused outrage for filming himself kissing a statue commemorating Korean wartime sex slaves has been sentenced to six months in prison, a court in Seoul said yesterday. Johnny Somali, 25, gained notoriety several years ago for recording himself doing a series of provocative stunts in South Korea and Japan, and streaming them on platforms such as YouTube and Twitch. South Korean authorities indicted Somali — whose real name is Ramsey Khalid Ismael — in 2024 on public order violations and obstruction of business, and banned him from leaving the country. “The court has sentenced him to six months in
The death toll from a shooting in western Afghanistan rose to 11 on Saturday, after gunmen targeted civilians at a picnic spot in Herat, the provincial authority said. Bullet marks were visible on a wall of the Sayed Mohammad Agha Shia shrine, while bloodstains marked a blanket abandoned at the scene. “Eleven people have been recorded dead and eight others wounded from Friday’s incident, with the condition of two of the wounded reported as critical,” Herat’s information office said in a statement. The update raises a toll of seven killed provided on Friday by the Afghan Ministry of Interior Affairs
RSS