Image from Freepik
AI-Powered Robotic Rats Successfully Blend Into Real Rat Communities
- Written by Kiara Fabbri Former Tech News Writer
- Fact-Checked by Justyn Newman Former Lead Cybersecurity Editor
Roboticists at the Beijing Institute of Technology, in partnership with colleagues from the Technical University of Munich, have developed a robotic rat capable of social interaction with live rats for a continuous half hour.
In a Rush? Here are the Quick Facts!
- Robotic rats mimic social behaviors, fooling real rats into accepting them during interactions.
- AI-driven training taught robots to wrestle, nuzzle, and display aggression like real rats.
- Robotic rats enable long-term studies of social dynamics, overcoming biological limitations.
In a study published in Nature Machine Intelligence , the team details how AI was utilized to train the robot to replicate real rat behavior.
Tech Xplore reports that Thomas Schmickl from the University of Graz explained how the researchers employed feedback loops and AI-based reinforcement learning to equip the robotic rats with social skills convincing enough to engage live rats.
The goal of the project was to create a robot that could convincingly mimic the behavior of a real rat to the point that lab rats would accept it as one of their own. It appears that the researchers have succeeded.
Studies have shown that rats display a range of social behaviors, from aggression during stress to playfulness when content.
For the robot to be accepted, it had to accurately perform these behaviors, including wrestling and nuzzling during positive interactions or showing aggression when appropriate, as reported by Tech Xplore.
To achieve this, the team designed the robotic rat with AI-driven deep learning capabilities and trained it using video recordings of rats interacting.
To enhance the robot’s social acceptance, the researchers coated it with rat urine, which masked its artificial scent and made it seem more familiar and less threatening to the real rats, as reported by Interesting Engineering .
Over time, the robot refined its responses and learned to behave appropriately in real-time interactions with rats, receiving positive reinforcement when it succeeded. This allowed it to adapt and develop what could be described as a “rat personality,” noted Tech Xplore.
While the robot differed from a biological rat in having a wheeled, cart-like body instead of legs, it retained several lifelike characteristics, reports Tech Xplore. Its spine could bend and twist like a real rat’s, its head movements were realistic, and its forelimbs were functional enough to physically engage with real rats.
Tests demonstrated that the robot was not only accepted but elicited expected responses. Real rats reacted fearfully when the robot displayed aggression and engaged in playful wrestling or nuzzling during calm moments, just as they would with other rats, as reported by Tech Xplore.
The researchers suggest these robots could be valuable tools for studying social dynamics and influencing the emotional states of lab rats.
The researchers highlight that one key advantage of these robotic rats lies in their ability to engage in long-term and repetitive interactions, overcoming the inherent limitations of natural social behaviors in biological systems.
This extended interaction capability opens new opportunities for conducting more in-depth and prolonged studies on social dynamics and behavioral patterns. The team believes that this particular robotic rat, equipped with advanced AI training, represents a significant leap forward in robot-animal interaction.
Its ability to adapt to various scenarios positions it as a valuable tool for studying animal behavior and potentially improving the welfare of laboratory animals, as reported by Interesting Engineering.
Image by Freepik
Encrypted Cyber Attacks Surge: 87% of Threats Now Hidden In HTTPS Traffic
- Written by Kiara Fabbri Former Tech News Writer
- Fact-Checked by Justyn Newman Former Lead Cybersecurity Editor
Recent research by cloud security firm Zscaler highlights the growing use of encryption by cybercriminals.
In a Rush? Here are the Quick Facts!
- 87% of cyber threats now use encrypted channels, up 10% from last year.
- Malware accounts for 86% of encrypted attacks, totaling 27.8 billion incidents.
- The manufacturing sector faced 42% of encrypted attacks, the highest among industries.
The widespread adoption of HTTPS encryption across the internet has created new challenges for cybersecurity teams.
While encryption safeguards legitimate traffic, it also enables malicious actors to hide their activities from traditional security tools, complicating the balance between data privacy and threat detection.
The study by Zscaler revealed that 87% of cyber threats now utilize encrypted channels, a 10% increase from the previous year, as reported by Cyber Magazine .
These findings, derived from the analysis of 32.1 billion blocked threats between October 2023 and September 2024, underscore how attackers are exploiting HTTPS protocols to evade detection, says Cyber Magazine.
The rise in encrypted attacks coincides with increased adoption of cloud services and remote work solutions, which expand the attack surface for organizations.
Traditional security methods struggle to inspect encrypted traffic at scale, leaving potential blind spots in enterprise defenses, notes Cyber Magazine.
“The rise in encrypted attacks is a real concern as a significant share of threats are now delivered over HTTPS,” said Deepen Desai, Chief Security Officer at Zscaler, as reported by Cyber Magazine.
He emphasized the importance of implementing zero trust architecture and large-scale TLS/SSL inspection to counter these sophisticated attacks. Cyber Magazine reports that the manufacturing sector emerged as the most targeted industry, with 42% of detected encrypted attacks.
This marks a 44% year-on-year increase, driven by the sector’s adoption of connected systems and Industry 4.0 technologies. Technology, communications, services, education, and retail sectors also faced significant threats.
Geographically, the United States recorded the highest number of encrypted attacks, at 11 billion incidents, followed by India with 5.4 billion. France, the United Kingdom, and Australia were also among the most affected nations, as reported by Cyber Magazine.
Malware accounted for 86% of encrypted attacks, with notable variants such as AsyncRAT and Atomic Stealer exploiting encrypted channels, reports Cyber Magazine.
Web-based attacks surged, with cryptomining up 123% and phishing increasing by 34%. Researchers suggest AI tools may be fueling this growth , reports Cyber Magazine.
Zscaler recommends full TLS/SSL inspection and microsegmentation as critical defenses. Its Zero Trust Exchange platform offers security measures across all stages of an attack, from reconnaissance to data exfiltration.
“With threat actors focused on exploiting encrypted channels to deliver advanced threats and exfiltrate data, organisations must implement a zero trust architecture,” Desai added, as reported by Cyber Magazine.
