Chrome for Developers’ Post

If you’re involved in Ai, you should keep this link saved and reference it often as an evaluator of the technology. With that said, in the enterprise, should you be using technology that has 700+ types of IDENTIFIED risks? I’ll let you be your own judge, however, you should consider alternatives where available no matter what an “evangelist” tells you.

#ISO42005 (AI Impact Assessment Standard) rolls all of these reviews into one, and suggests that they need to be collaborative and documented to do effectively, multiple times over the life cycle of AI. AI impact assessments include, but are not limited to, #legal, #risk, #privacy, #accessibility #cyber #ESG and more and fundamentally, commercial questions about cost/benefit if all the other thresholds meet the risk appetite of the Board. Having done exactly this for the past few years, I can say that without coordination of subject matter experts, reviews are fragmented and lack systems thinking to effectively assess risks and capture benefits. Risk alone is very important, but companies that have check-lists for AI risk miss the bigger picture and commercial opportunities. All technology is a trade-off. Those trade-offs need to be accepted by execs who carry the responsibility for the decisions, subject to advise by multiple subject matter experts. Smaller companies are going to need to rely on advisory boards with deep expertise for this work.

"With our zero trust strategy, plus the Workers AI platform, you can get high-performing applications out of our network with all the tool sets you need." As zero-trust security models become more prevalent across IT teams, it's imperative that the tooling enables a seamless integration process to keep things running smoothly, efficiently, and, most importantly, securely. At Cloudflare, the company enables its customers to harness zero trust while utilizing the power of AI, enabling a streamlined and secure posture across connectivity environments. On this week's episode of ESI, Mike Hamilton, CIO of Cloudflare, provides an inside look into how the company utilizes AI to enable a more robust user experience. Full episode here: https://lnkd.in/giBG7Kg3 #esipodcast #cioinsights #ciopodcast #zerotrust #aitools #enterpriseai #ai #ml #aiapplications #aipodcast

MIT just unveiled the most impressive tool for #AIgovernance I've seen to date: https://airisk.mit.edu. ▶ Its Risk Database identifies 700+ types of risks captured from over 43 frameworks (with attribution). ▶ Its Casual Taxonomy of AI Risks classifies how, when, and why these risks occur. ▶ Its Domain Taxonomy of AI Risks places these risks into 7 domains and 23 subdomains. This has so many potential applications, from targeting research topics (e.g., for mitigation) to developing audit protocols to building learning paths to shaping policy. Here's the short explainer video: https://lnkd.in/ejD7MvwB. Can't wait to dig into this.

It is a very powerful asset to look at AI projects.

Thanks to Kevin Fumai for raising awareness for this new tool from #MIT. For those of you in the initial stages of building your AIMS, please do not discount the value of this resource because of it's cost (free). For those of you with semi-mature AIMS, please evaluate for utility in your current AI Risk Management processes. Shea Brown BABL AI Walter Haydock StackAware Christian Hyatt risk3sixty Jacob Nix RISCPoint Chris Arrendale CyberData Pros A-LIGN #AIRisk #ISO42001 #AIGovernance #TheBusinessofCompliance #ComplianceAlignedtoYou

"What is the AI Risk Repository? The AI Risk Repository has three parts: 1) The AI Risk Database captures 700+ risks extracted from 43 existing frameworks, with quotes and page numbers. 2) The Causal Taxonomy of AI Risks classifies how, when, and why these risks occur. 3) The Domain Taxonomy of AI Risks classifies these risks into seven domains (e.g., “Misinformation”) and 23 subdomains (e.g., “False or misleading information”)." https://airisk.mit.edu

The more I understand the potential of AI in various sectors the more I become aware of it’s undermining risks and I couldn’t agree more with the following “Lack of shared understanding of AI risks can impede our ability to comprehensively discuss, research, and react to them.” This paper addresses this gap by creating an AI Risk Repository to serve as a common frame of reference.

Fantastic resource for all AI risk and taxonomy lovers! MIT FutureTech made it possible : meet "The AI Risk Repository: A Comprehensive Meta-Review, Database, and Taxonomy of Risks From Artificial Intelligence" AI risks are a significant concern across various stakeholders, with over 3,000 real-world incidents documented in the AI Incident Database. Researchers have attempted to classify these risks, but efforts have been uncoordinated, leading to conflicting classification systems. The inconsistent use of terms and concepts across different classification systems makes it challenging to understand the full scope of AI risks. This work aims to create a comprehensive, unified classification system by synthesizing diverse perspectives from previous efforts. The result is two distinct classification systems: one for high-level causes of AI risks and another for mid-level hazards or harms from AI, which together form a unified approach to understanding AI risks. See full repository, explanation video and access to database right here : https://airisk.mit.edu/