← Back

Wire · regulatory

AI developers to do more to stop growing terrorist, extremist use

Published

16 July 2026

Topic

regulatory

Sectors

AI Frontier ModelsGenerative AI

Geography

Australia

Source

Read at abc.net.au

Verified

Fusion42 · 16 July 2026 · Fusion42 review

Tech Against Terrorism report finds AI models like ChatGPT fail to block 52% of requests for terrorist information; safeguards are easily circumvented, particularly in open-source models. Australian authorities flag extremist use of AI for attack planning as a major national security concern, with documented cases linked to 70+ deaths.

This Wire brief sits within Fusion42's coverage of AI Frontier Models and Generative AI.

◆ The Wire takeaway

If you're building or deploying AI infrastructure, governments are about to mandate what you can't let users do with it - and the bar is rising faster than your safeguards are. The next 18 months will likely see legislation that treats failure to block weaponisation requests as corporate negligence, not a feature request.

Coverage

1 source · 16 Jul 2026

Related on Wire

Topics

AI Frontier ModelsGenerative AIai-safetyextremismnational-securitycontent-moderationregulatory-pressure