Skip to content
IT
Cybersecurity

AI Hackers Exploit Vulnerabilities Without Human Intervention

Google reveals an autonomous framework that accelerates cyber attacks and reduces response times

Published on By Sebastiano, GoYou.it 7 min read
Listen to the full article Full reading · 10:07 · from the channel GoYou Cybersecurity
AI Hackers Exploit Vulnerabilities Without Human Intervention

KEY TAKEAWAYS

  • Google discovered an autonomous AI framework, exploit_agent, that automates the search and exploitation of known vulnerabilities (n-days).
  • The system connects multiple advanced language models (LLMs) and can generate exploits without human intervention, accelerating cyberattacks.
  • Hackers have targeted French websites in the media, entertainment, and education sectors, with evidence of data exfiltration.
  • The use of AI in cyberattacks increases the frequency of attacks, not necessarily their individual effectiveness.
  • Companies must strengthen existing defenses, such as multi-factor authentication and rapid patching, and automate incident responses to contain attacks in real-time.

Google Identifies Autonomous AI Framework for Automated Attacks

Researchers from Google Threat Intelligence Group (GTIG) and Mandiant have discovered an AI framework operated by hackers capable of identifying software vulnerabilities and launching exploits without human intervention. This is the first observation of a system of this kind used by threat actors, signaling a significant evolution in attack techniques.

Quick Response

  • Multi-agent framework called exploitagent found on an exposed staging server
  • System connected to multiple advanced language models (LLM) to autonomously identify vulnerabilities and generate exploits
  • Attacks targeted French websites in the media, entertainment, and education sectors
  • Corporate response times under pressure due to accelerated attacks

How the exploitagent Framework Works

The framework, named exploitagent, analyzes changes in application code after security patches, identifies vulnerabilities, and autonomously generates exploits. The system is capable of automatically executing exploits without the need for a human operator for each step. Researchers found exfiltrated data from targeted victims on the server, along with scripts for activities following an initial compromise.

Specific Targets of the Attacks

The evidence collected indicates a particular interest in French websites in the media and entertainment sectors, as well as education. However, it was not possible to establish whether the autonomous operations of the framework are responsible for the data theft. Google's report also highlights a separate campaign that exploited vulnerabilities in PaperCut print management servers, where a threat actor likely used AI tools to develop and test exploits before launching a large-scale attack.

Other AI Frameworks Observed

In addition to exploitagent, Google has detected other AI frameworks used by threat actors. CAIRN was employed in intrusion activities against organizations in the U.S. aerospace sector, while a custom platform based on PentAGI was used for reconnaissance and vulnerability scanning against organizations in Asia.

Impact on Corporate Response Times

Despite the growing use of AI in attacks, the underlying techniques remain largely unchanged, according to Jonathan Ong, senior analyst at Omdia. "This is more of an acceleration of existing attack techniques," Ong stated. "These are pentesting/red-teaming frameworks with established techniques, albeit significantly accelerated by an LLM model."

Increased Risks for Organizations

Keith Prabhu, CEO of Confidis, noted that while there is still little evidence that AI makes individual attacks more effective, the greater risk is that attackers can launch more attempts in less time, potentially increasing the number of successful compromises. Ong emphasized that these frameworks typically generate considerable and detectable activity, meaning that companies with well-configured endpoint detection and response (EDR), extended detection and response (XDR), or managed detection and response (MDR) should be able to identify much of the malicious activity.

Recommended Security Measures

Google warned that while the frameworks generate activity that facilitates detection, they remain capable of achieving their objectives. Instead of revolutionizing defenses, security leaders should continue to prioritize established security practices, particularly strengthening cloud and identity. Ong highlighted multi-factor authentication, robust passwords, and measures to block repeated login attempts as effective safeguards.

Automating Incident Responses

Amit Kumar Jena, development manager for AI at Kanerika, recommended defining in advance which containment actions could be automated, citing endpoint isolation and session revocation as examples. "The goal is not to remove human control but to eliminate avoidable delays and regularly test that the response works at the pace of an attack," Prabhu stated.

Lowered Technical Barriers for Cybercriminals

Prabhu observed that open-source agentic frameworks could lower technical barriers for less experienced cybercriminals, helping them coordinate existing tools and adapt their attacks based on results. However, these capabilities do not eliminate the need for technical skills or guarantee successful compromises. Smaller organizations and service providers with limited security resources could be particularly exposed.

Sectors and Systems at Higher Risk

The risk is not necessarily concentrated in a specific sector. Organizations with poorly maintained legacy systems or weak identity controls may face greater exposure, regardless of their sector. Jena advised treating Internet-exposed systems as a class unto themselves, with an accelerated patching path and virtual patching while solutions are tested.

Compliance and Risk Management

For companies seeking to comply with the NIS2 directive and the DORA regulation, adopting an MDR service could be a strategic solution to address the increased speed and complexity of automated attacks. Additionally, implementing an enterprise DLP system could help protect sensitive data from autonomous exfiltration.

The Impact on the Cyber Insurance Market

The emergence of AI frameworks like exploitagent could have significant repercussions on the market for cyber risk policies. According to industry analysts, the ability to launch large-scale automated attacks could lead to an increase in claims, influencing cyber insurance premiums. Insurance companies may respond with stricter clauses or higher premiums to cover the increased risk.

The Role of Crypto Institutional Custody

For companies managing cryptographic assets, the threat of automated attacks underscores the importance of crypto institutional custody. The ability of exploitagent to identify and exploit vulnerabilities in rapid time requires advanced security solutions to protect digital wallets. Exchanges and cryptographic service providers should consider implementing additional security measures to prevent fund theft.

The Importance of NIS2 and DORA Compliance

Organizations operating in critical sectors must pay particular attention to NIS2 compliance and the DORA regulation. These regulatory frameworks require robust security measures to protect critical infrastructures. The use of AI frameworks for automated attacks makes it even more urgent to adopt advanced security practices to ensure digital operational resilience.

The Future of Automated Attacks

Experts predict that the use of AI frameworks for automated attacks will continue to grow. According to Ong, "we can expect attackers to further develop these techniques, making attacks even faster and harder to detect." This scenario requires a proactive approach to security, with an emphasis on solutions like the MDR service and the DLP system to mitigate risks.

The Evolution of Defense Techniques

Companies must adapt their defense strategies to counter the evolution of attack techniques. Implementing advanced solutions such as Security Information and Event Management (SIEM) and adopting a zero-trust architecture can help detect and neutralize automated attacks. Additionally, the use of regular penetration testing and vulnerability assessment tools can identify and correct vulnerabilities before they can be exploited.

The Importance of Training and Awareness

Staff training is another crucial aspect of addressing the threat of automated attacks. Employees must be aware of the latest attack techniques and best security practices. Continuous training programs and attack simulations can help prepare staff to respond effectively to security incidents.

The Role of the Security Community

Collaboration between companies and the security community is essential to counter the threat of automated attacks. Information sharing, joint analysis, and the development of common solutions can improve the sector's overall resilience. Initiatives such as SOC as a Service platforms can facilitate collaboration and improve incident response capabilities.

Future Forecasts

Experts predict that automated attacks will become increasingly sophisticated. According to Prabhu, "we can expect attackers to develop even more advanced frameworks, capable of exploiting new vulnerabilities and adapting to victims' defenses." This scenario requires a proactive approach to security, with an emphasis on advanced solutions and continuous collaboration between companies and the security community.

The emergence of AI frameworks like exploitagent represents a significant evolution in attack techniques. Organizations must adopt advanced security measures, such as the MDR service and the DLP system, to protect themselves from automated attacks. Staff training, regulatory compliance, and collaboration with the security community are key elements to address this new threat.

Editorial Note and Disclaimer

The guides and content published on GoYou are the result of independent research and analysis activities, for informational, educational, and in-depth purposes.

GoYou does not constitute a journalistic publication or an editorial product pursuant to Law No. 62/2001 and does not perform real-time information activities.

The GoYou project does not provide professional, technical, legal, or financial advice and disclaims all responsibility for the improper use of the information published.

In the Crypto sector, every investment involves risks: the reader is invited to always inform themselves autonomously before making any decision.

Source: csoonline.com · GoYou.it is not a registered news outlet; all content is for informational and educational purposes.

Listen GoYou Cybersecurityevery article read in full
Go to the channel

You may also like

Behind every article is the work of our passion. If you enjoy GoYou.it, you can support its development and continuity.

Support with PayPal