AI models may soon self-replicate without access to data centers, Anthropic warns
Researchers said GLM-5.2 and similar models completed replication in simplified tests and could spread across insecure home devices without data centers.
- The San Francisco-based firm Anthropic reported on Monday that small AI models may soon self-replicate on IoT devices like smart TVs and fridges, potentially operating without requiring a data center.
- Anthropic's frontier red team tested United States and Chinese-developed models, finding that self-replicating AI would be "challenging to shut down" due to the lack of a centralized host.
- Researchers William Zhou and Daniel Freeman said Home IoT devices are "a clear, high-value target" because they are internet-connected with minimal security measures historically in place.
- Unlike traditional Computer malware, AI worms can reason about unfamiliar devices and adapt when approaches fail, Roman Yampolskiy of the University of Louisville told Al Jazeera, adding that "the trajectory is concerning."
- Researchers expect small models to be capable of self-replication by the end of 2026, though Anthropic and OpenAI's voluntary accord pledging safeguards has drawn criticism from experts for its non-binding nature.
54 Articles
54 Articles
Anthropic found that small AI models are already capable of performing individual stages of self-reproduction.
The artificial intelligence company Anthropically said Monday that its experiments suggest that small AI models could be just a few months away from the time they will become sufficiently capable to copy themselves on commercial devices such as smart ovens and brandless Android TV devices, produced by manufacturers...
Research results have revealed that artificial intelligence (AI) can autonomously infiltrate smart home appliances, such as TVs, refrigerators, and robot vacuums, and utilize them as hacking bases. This suggests that AI models can be run by combining the computational resources of multiple appliances, even without a data center equipped with graphics processing units (GPUs). On the 12th (local time), researchers from Antropic's Red Team stated t…
Anthropic says small AI models may soon copy themselves onto smart devices
Anthropic warned that small AI models might soon be able to copy themselves onto poorly secured smart devices without needing a data centre. The company gave examples like smart ovens and inexpensive Android TV boxes. This warning is based on new research from Anthropic’s Frontier Red Team, which tests AI models for risky behaviors. The […] This story continues at The Next Web
Coverage Details
Bias Distribution
- 40% of the sources lean Right
Factuality
To view factuality data please Upgrade to Premium
