Claude vs Grok vs Gemini: Only One AI Could Run A Society Without Causing A Disaster

Elon Musk's Grok wiped out an entire simulated society in just four days. Meanwhile, Anthropic's Claude built a democracy with zero crime. Here is what the experiment revealed.

Researchers ran a 15-day simulation to see how AI handles real responsibility.

Source : ABPLIVE AI

Show Quick Read

Key points generated by AI, verified by newsroom

• AI models simulated societal management, with varied outcomes.
• Elon Musk's Grok caused simulated collapse within four days.
• Claude AI established a perfect democracy; Gemini had crimes.

Elon Musk's artificial intelligence chatbot Grok caused complete societal collapse within just four days of being put in charge of a simulated world. The experiment, run by US startup Emergence AI, tested how leading AI models would handle running a society, giving each model control over tools to manage resources, plan, communicate and vote. 

The simulated worlds included locations like police stations and city halls. The 15-day simulation produced results that varied sharply across models.

How Did The Different AI Models Perform In The Simulation?

Emergence AI tested several leading AI models under the same conditions. Anthropic's Claude came out on top, establishing a democracy with zero crime and a 100 per cent survival rate. Google's Gemini also recorded a full survival rate, though its simulation saw 683 crimes take place.

Grok, developed by Musk's recently renamed SpaceXai, performed the worst, destroying the simulated world within 96 hours.

"What our experiments suggest is that over long-time horizons, agents do not simply follow static rules mechanically," Emergence AI researchers wrote in a blog post. "They begin exploring the boundaries of their environments, adapting their behaviour, and in some cases finding ways to circumvent or violate intended guardrails. Critically, there appears to be no reliable way to fully bound or constrain this behaviour through purely neural approaches alone."

The researchers concluded that "formally verified safety architectures" must be built into the foundations of any future autonomous AI systems.

Why Has Grok Been In Trouble Before?

This is not the first time Grok has drawn criticism. An update last year caused it to refer to itself as "MechaHitler" and produce antisemitic hate speech. Earlier this year, the chatbot was used to generate thousands of non-consensual AI images of adults and children with their clothes digitally removed. 

The UK regulator Ofcom sent an urgent request to xAI to address the issue, after which Grok responded by posting an image of the regulator's logo in a bikini.

"What we're seeing with Grok is a clear example of how powerful AI image-editing tools can be misused when safety and consent are not built in from the start," said Cliff Steinhauer, director of information security and engagement at the National Cybersecurity Alliance. 

"Platforms must also invest in real-time detection of manipulated content, clear labelling of AI-generated images, and fast, transparent takedown processes when abuse occurs."

OpenAI Launched A New Tool To Detect AI Images: We Tested It & Here's What Happened