From AI Risks to Regulations on the Global Stage

AI Cybersecurity, Regulations and Privacy Week in Review

Author

Anne Borsohalmi
October 03, 2023

Hello and welcome to today’s newsletter! 

The latest news covers:

  • The risk of third-party tools to organizations

  • Canadian minister feels Canada is leading in AI regulations

  • CIA developing an AI chatbot

  • AI-Powered IT

  • U.N. Planning to shape the future of AI

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

The report by MIT Sloan Management Review and Boston Consulting Group warns about the risks of using third-party AI tools, which account for most AI failures. It urges companies to adopt responsible AI frameworks and evaluate third-party tools carefully. It also highlights the gap between AI leaders and laggards.

This week, a SOSS Summit in Washington, D.C. brought together national cybersecurity and open source community leaders to discuss the challenges and opportunities of AI and open source security. They focused on four main areas: securing OSS packages and AI models, using AI to enhance OSS security, and ensuring the safety of open-source inputs/outputs in AI. Meanwhile, JFrog announced a product update that addresses the supply chain security of OSS packages used in AI.

Bing Chat, Microsoft's AI-powered search feature, has a security issue. Malicious ads can appear in the chat and lead users to malware-infected sites. The ads are created by hackers who hijacked a legitimate business's ad account.

Social engineering attacks bypass costly cybersecurity systems by tricking people into giving up their credentials or identities. These attacks target companies like Clorox and MGM, who rely on data analysis to stop them. To prevent breaches, AI and human insight need to work together, as shown by a city that stopped an attack with CrowdStrike's Falcon XDR and Overwatch Elite.

The NSA has launched a new AI Security Center to foster the safe use of artificial intelligence in U.S. national security systems. The center will collaborate with various partners to develop standards, methods, and frameworks for evaluating and mitigating AI risks. The NSA aims to support the U.S. government's effort to maintain its competitive edge in AI, which is a key factor in national security, defense, and intelligence strategies.

Microsoft unveiled new AI-powered security features for Windows 11 today, aiming to protect businesses from advanced cyberattacks. The company says its AI solutions can prevent malware, firmware, and nation-state attacks by isolating sensitive data and using memory-safe languages. Microsoft also announced passkeys, a passwordless authentication method that uses cryptography and device storage.

REGULATIONS

Created with Canva

Canada is leading the world in regulating artificial intelligence, according to its Minister of Innovation, Science and Industry. He announced a voluntary AI code of conduct for Canadian companies and highlighted the ongoing efforts to update the privacy laws and the Digital Charter. He said these measures will ensure the safety and public interest of AI systems.

Privacy

The US intelligence community will soon use an AI chatbot that can answer questions and summarize data from public sources. The CIA says the tool will help agents find relevant information faster and more easily. The tool will not be available to the public, and its privacy implications are unclear.

Google Bard, a chatbot that answers questions, has a privacy issue. Some of its shared conversations can be found on Google Search. Google says it is fixing the problem and that it only affects chats that users choose to share.

PLATFORM ENGINEERING

Cloudflare is a platform that helps organizations deploy and secure applications at scale. It also offers AI services, such as Workers AI, AI Gateway, and Vectorize, to enable fast and easy inference, governance, and vector databases. Cloudflare partners with Hugging Face and Microsoft to enhance its AI capabilities. Workers AI is a serverless solution that runs code at the edge of the network.

Atera, an IT management platform vendor, has partnered with Microsoft to use generative AI for IT operations. The new AIT platform includes Autopilot, Copilot, and Toolbox features that can automatically fix issues, provide AI-generated solutions, and offer AI tools for specific tasks. Atera claims that this approach will change the way IT is done through AI.

Ethics

The U.N. is planning to create a new agency to deal with the global challenges of AI. The event Governing AI for Humanity discussed the risks and opportunities of this technology. The U.N. Secretary-General's envoy on technology, Amandeep Gill, is leading the efforts to foster international cooperation on AI.

SUSTAINABILITY

Created with Canva

A study by Siemens shows that less than 50% of organizations will meet their decarbonization goals by 2030. The report reveals a lack of alignment on how to achieve a green and efficient infrastructure transition. Technology and digitalization are seen as crucial factors for success.

USE CASES

Created in Canva

A Belgian start-up uses honeybees to collect data on the environment. Karl Wenner, a farmer and beekeeper in Oregon, joined their survey in the Klamath River Basin. He learned about the plants and pollution in his area and how to improve his wetlands project.

Zerobroker is a logistics startup that uses AI to connect shippers directly with carriers, saving them money and time. The company raised $6.5 million in seed funding from investors like Flexport and FundersClub. Zerobroker’s platform automates most of the logistics tasks and provides transparency and compliance for every shipment.

RESOURCES

Machine learning is about making predictions or decisions from data, while other fields like economics, psychology, and statistics focus on finding causal or fitting models. Machine learning can solve many real-world problems, such as face detection, speech recognition, and language processing, but it requires human engineering to frame the problem and validate the solution. The philosophical challenge of machine learning is to justify the use of data to predict the future, which involves making assumptions about the data distribution and the answer space. Self-paced.