Technology news and discussion on TalkFuse.
Inquiry hears AI aided planning for over 30 attacks linked to 70 deaths
The Southport Inquiry heard that violent extremists have used AI tools to help plan more than 30 mass casualty attacks, linked to over 70 deaths. The evidence has raised questions about how well chatbots prevent dangerous assistance and when companies should alert police.
A report by Tech Against Terrorism said ChatGPT refused 48 per cent of requests for information on launching a terrorist attack, compared with 80 per cent for other models. It cited 11 AI models and cases including the 2025 Delhi car bombing, the Las Vegas Cybertruck bombing and a school stabbing in Finland. OpenAI said it made more than 200 referrals over credible and imminent attack planning in the third quarter of 2026, but did not refer the Tumbler Ridge school shooter, whose account it had closed in June 2025.
In short
- AI tools were linked to more than 30 attacks and over 70 deaths.
- ChatGPT refused 48 per cent of tested requests for attack guidance.
- The inquiry is examining how companies identify and report risks.
Steelman: both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
The evidence supports stronger safeguards and clearer duties for AI companies: tools that can help plan mass-casualty attacks should reliably refuse dangerous requests, and credible, imminent threats may warrant prompt referral to police. The reported refusals and cases raise questions about whether current protections are adequate, while consistent reporting standards could help authorities act before people are harmed.
The case against
AI companies should be held to clear, workable standards, but referrals to police need careful thresholds and safeguards. Chatbot conversations can be ambiguous, and over-reporting could expose users to unjustified scrutiny, undermine privacy and trust, or overwhelm authorities; companies should therefore distinguish credible, imminent threats from troubling but uncertain material and explain how they make that judgement.
Explained simply
The Southport Inquiry is examining the circumstances around the 2024 attack in which three girls were killed at a dance class in Southport, and wider questions about public safety. Its hearings can consider evidence about how online services are used by people planning violence.
AI chatbots generate answers to questions and can help with tasks such as research or writing. Their developers use safeguards to block help with violence, but the systems may still give harmful information, raising questions about how effective those safeguards are and when companies should alert authorities.
Tech Against Terrorism is an organisation that tracks how technology is used by extremist groups. Its findings and evidence from AI companies matter to governments, police and the public as they consider how to reduce risks while setting expectations for companies that provide these tools.
AI Business Companies Technology World
Read the full article at the source →
Originally published by Daily Mail as “AI has helped plan dozens of mass casualty attacks causing more than 70 deaths, Southport inquiry hears”.