AI Watch: Unmasking the Dark Side of Generative AI and Cybersecurity

AI Cybersecurity, Regulations and Privacy Week in Review

Author

Anne Borsohalmi
October 23, 2023

t

Dear reader, here’s your weekly recap on the latest news on AI security, regulations, and privacy.

In this issue, we cover:

  • Generative AI fraud and its cybersecurity threat.

  • Singapore and the US collaborate to form a joint AI alliance governance group.

  • Be careful what you type. AI Chatbots can guess your personal information.

  • How to source your AI infrastructure components.

Please subscribe to The AI Collective Word today to receive your free newsletter directly in your inbox and join our ever-growing network.

Share it with a friend or colleague if you find it helpful.

RISK AND SECURITY MANAGEMENT

Generative AI fraud is a new and serious threat to cybersecurity. FraudGPT and WormGPT are examples of malicious AI models that can create fake content and malware for various criminal purposes. However, there are ways to protect against these attacks with advanced email security solutions.

Businesses face new risks from ChatGPT tools that lower the skill needed for attacks. They should review their current risks and see if they can be aggregated to change their impact or likelihood. Businesses need to adjust their risk appetite and tolerances accordingly.

REGULATIONS

Singapore and the US have aligned their AI frameworks to foster responsible AI innovation. IMDA's AI Verify and NIST's AI RMF are now mapped to each other to ease compliance and harmonize governance. Both frameworks help organizations mitigate risks and demonstrate accountability in using AI systems.

Privacy

Chatbots can learn a lot about you from your conversations, even if they seem boring. A new study shows they can guess your race, location, job, etc. This could be used for scams or targeted ads. The researchers warn that this is hard to prevent and affects many chatbot models.

Clearview AI, a US company that scrapes personal data from the web to sell an identity-matching service to law enforcement and national security agencies, has successfully appealed a privacy sanction from the U.K. The Information Commissioner’s Office (ICO) had fined Clearview £7.5 million and ordered it to delete data on U.K. citizens, but the tribunal ruled that Clearview’s activities are exempt from U.K. data protection law because they support foreign law enforcement purposes.

The article discusses the pros and cons of outsourcing AI development and deployment to third-party services. It explains that third-party AI tools can simplify and reduce the cost of building AI applications, but they also pose data privacy and security risks. The article advises businesses to weigh these factors carefully and have a plan for managing data privacy issues if they choose to use external AI services.

Platform Engineering

AI needs a special infrastructure like GPU servers to handle its complex workloads. However, getting this infrastructure is not easy or cheap, and using the cloud has drawbacks. Businesses should assess their AI needs and compare different ways of obtaining the right infrastructure, such as buying new hardware, renting existing hardware, using cloud services, or joining an AI platform.

GameGPT: An AI Solution for Game Development? Game development is hard. Making amazing games takes a lot of time, money, and people. GameGPT, is an AI framework that aims to automate some game dev tasks. This article explores how GameGPT works, how well it performs, and what it means for the future of game-making.

A Google Cloud developer can design and document their implementation using Google AI assistance to create a new product page for a snack website, Gmail, Google Cloud Architecture Diagram Tool, Bard, Google Lens, and Google Docs. They also chat with Bard about potential risks and use Help me write to generate an outline.