TechRisk #76: Microsoft, Nvidia, OpenAI's first AI cyber attack simulation with Fed

Plus, Microsoft prioritizes security over AI tech, DeFi UwU hacked again and lost $23M, Terraform Lab shuts down and more!

Photo by Ümit Yıldırım on Unsplash

Tech Risk Reading Picks

1. Fed and tech companies’ first AI attack simulation: The Cybersecurity and Infrastructure Security Agency (CISA) recently conducted its inaugural tabletop exercise, engaging over 50 AI experts from government and industry. Led by the Joint Cyber Defense Collaborative (JCDC), the exercise simulated a cybersecurity incident targeting an AI-enabled system. Participants practiced incident response strategies across three modules, focusing on mitigating damage caused by a hypothetical attack that bypassed an organization’s AI defenses in its email system. Major industry players such as Amazon Web Services, Cisco, IBM, Microsoft, and NVIDIA, alongside government agencies like the FBI and NSA, participated in the exercise. OpenAI's Head of Security, Matt Knight, underscored the collaborative nature of cybersecurity efforts and OpenAI's commitment to safe AI development. Overall, the tabletop exercise signifies a proactive approach to understanding and mitigating cyber threats to AI systems through collaboration between government, industry, and security experts. [more] [more-2]

   1. Here are the key takeaways:

      1. Unique Challenges of AI-Enabled Systems: Traditional cybersecurity strategies may not suffice against attacks targeting AI systems. As AI tools become more prevalent, hackers could exploit them to accelerate and scale their attacks.

      2. Collaborative Approach: The tabletop exercise hosted by CISA at Microsoft's offices involved over 50 AI experts from various sectors. Participants included major players like Amazon Web Services, Microsoft, Nvidia, OpenAI, and Palantir. This collaborative effort aimed to simulate real-world scenarios and refine response strategies.

      3. Identifying and Addressing Threats: The exercise focused on current threats and potential future attack vectors leveraging AI. Participants discussed how to effectively communicate and respond to incidents involving AI systems, emphasizing the need for clear protocols between government and private sectors.

      4. Preparation and Playbook Development: Insights from the simulation will inform the development of CISA's AI security incident playbook, expected to be released by year-end. This playbook aims to standardize responses to AI-related cyber incidents, ensuring a coordinated and effective approach.

      5. Future Preparedness: The JCDC plans to conduct additional tabletop exercises to further refine strategies and anticipate new threats. This ongoing effort underscores the proactive stance of both government and industry in mitigating risks associated with AI technologies.

2. Microsoft prioritizes security over AI tech: Microsoft is undergoing a significant cultural shift towards prioritizing cybersecurity following several major security breaches. Brad Smith testified before Congress, acknowledging past failures and outlining Microsoft's commitment to enhancing security measures. This includes CEO Satya Nadella taking personal accountability for security, tying executive salaries to security goals, and adopting all recommendations from the Cyber Safety Review Board. Microsoft aims to embed security in all aspects of its operations, investing heavily in cybersecurity resources and expanding its security team. This initiative marks a proactive effort to rebuild trust and mitigate future cybersecurity risks. [more]

3. Pope on AI risk: Pope Francis addressed the G7 Summit, urging leaders to prioritize human dignity in the development and use of artificial intelligence (AI). He warned against reducing human relations to algorithms and emphasized that decisions impacting people's lives should always be made by humans, not machines. The Pope also called for a ban on autonomous weapons, arguing that machines should never have the power to choose to take a human life. His ethical stance influenced the G7's final statement, which committed to regulating AI to ensure it remains human-centered, respects human rights, and supports sustainable development. [more]

4. Small businesses targeted by AI-drive threats: Recent surveys show a significant rise in ransomware attacks, particularly targeting small businesses, highlighting their vulnerability to AI-driven cyber threats. Oxford Capital points out key concerns for 2024, including, i) AI-powered phishing attacks deceive with personalized emails. ii) Automated exploits target software vulnerabilities swiftly. iii) Deep fake technology impersonates executives for fraud. iv) AI-driven ransomware encrypts data rapidly. v) Malicious AI bots exploit weak security measures. vi) Weak passwords pose a significant risk. [more]

   1. Small businesses can protect themselves with strong passwords, multi-factor authentication, software updates, employee training, and robust security solutions. These measures are critical as AI continues to transform cyber threats.

5. Hybrid Cloud security risk: In Gigamon's latest 2024 Hybrid Cloud Security Survey, it highlights the evolving nature of cyber threats, the complexities introduced by cloud environments, and the pressing need for organizations to adapt their cybersecurity strategies to effectively protect against sophisticated attacks. [more]

   1. Increase in Undetected Breaches: The survey reports a significant 20% year-on-year rise in undetected breaches. This highlights a growing challenge for organizations in detecting and responding to cyber threats effectively.

   2. Cloud Complexity: The diffusion of data across on-premises, public cloud, and private cloud environments is cited as a major hurdle. This complexity makes it increasingly difficult for organizations to maintain comprehensive visibility and control over their data, thus contributing to the rise in undetected breaches.

   3. Detection and Response Capabilities Decline: Compared to the 2023 report, there has been a decline in organizations' abilities to detect and respond to breaches. Only one in three organizations were able to detect a breach in the last year, and just 25% could respond in real-time.

   4. Visibility Challenges: There's a notable decrease in visibility into East-West (lateral) traffic, a critical aspect of cloud security. Despite its importance, only 40% of respondents reported having visibility into this traffic flow, down from 48% in the previous year.

   5. Perception of Cyber Risk: A significant 83% of respondents believe that cloud complexity increases their cyber risk. This sentiment is compounded by concerns over AI-powered cyber attacks, with 82% predicting an increase in ransomware threats globally.

   6. Tooling Insufficiency: Many organizations acknowledge that their current security tooling falls short. Sixty-five percent of respondents believe their existing tools cannot effectively detect breaches.

   7. Security Challenges in Hybrid Cloud Environments: Securing hybrid cloud environments poses several challenges, including ensuring data security and compliance across different cloud platforms, managing identities and access consistently, and securing East-West traffic between workloads.

   8. Future Security Investments: Despite these challenges, information security spending is projected to reach $215 billion in 2024, underscoring the urgency and priority placed on improving cybersecurity measures.

   9. Importance of Deep Observability: Key stakeholders emphasize the critical need for comprehensive visibility and deep observability to combat modern cyber threats effectively. This includes integrating AI-powered solutions, adopting Zero Trust frameworks, and enhancing detection and response capabilities.

6. Web3 Cryptospace Spotlight

   1. Kraken’s fake whitehat exploitation: A security researcher found a bug in Kraken's system and used it to withdraw over $3 million in digital assets. They demanded a reward from Kraken and refused to return the funds until certain conditions were met. Kraken emphasized no user funds were affected and is working with law enforcement to recover the stolen assets. This incident highlights ethical concerns in cryptocurrency security practices. [more]

   2. Terraform Labs shutdown: Chris Amani, CEO of Terraform Labs, has announced that the company will shut down after reaching a $4.47 billion settlement with the SEC. Terraform Labs plans to sell off important projects in the Terra ecosystem. Additionally, they intend to hand over control of the Terra blockchain to the community. [more]

   3. UwU Lend gotten attack again: DeFi lending protocol UwU Lend has suffered two attacks in the past three days. The second exploit occurred on Thursday during the protocol’s reimbursement process from the first hack. The ongoing saga has taken around $23 million from the protocol. [more]