news.com.au
Details
- Date Published
- 4 July 2026
- Priority Score
- 3
- Australian
- Yes
- Created
- 5 July 2026, 02:00 am
Authors (1)
- Theo HourmouzisENRICHED
Description
The Royal Commission on Antisemitism and Social Cohesion has heard from a major tech company about how artificial intelligence could be used to curb the spread of abusive material online. Anthropic says its AI system, Claude, includes safeguards designed to detect and stop hate speech. The company’s Australian general manager, Theo Hourmouzis, says in cases involving extreme violence, the platform may also alert law enforcement.
Summary
Anthropic's testimony to the Australian Royal Commission on Antisemitism highlights the potential for frontier AI systems like Claude to proactively moderate and intercept extreme online harms. The discussion emphasizes the integration of safety safeguards within large language models to detect hate speech and facilitate law enforcement alerts in cases of extreme violence. These developments illustrate the intersection of Australian judicial inquiries and global AI safety standards, specifically regarding the mitigation of societal-wide impacts of misinformation and radicalization. This context is critical for understanding how frontier capabilities are being repurposed as defensive tools within national governance frameworks to reduce catastrophic social contagion risks.