Galym Uteulin’s Post

The Launch of AISIC (a US AI regulatory body) Following some big news in the EU, in a strategic move to address the burgeoning challenges of artificial intelligence (AI), the U.S. government, under the Biden-Harris Administration, has unveiled the U.S. AI Safety Institute Consortium (AISIC). This initiative brings together over 200 stakeholders, including tech giants like OpenAI, Google, and Microsoft, to foster the safe development and deployment of AI technologies. 𝑵𝒐𝒕𝒆: 𝑻𝒉𝒆𝒚 𝒉𝒂𝒗𝒆𝒏'𝒕 𝒚𝒆𝒕 𝒂𝒔𝒌𝒆𝒅 𝑯𝑸 𝒕𝒐 𝒃𝒆 𝒂𝒕 𝒕𝒉𝒆 𝒕𝒂𝒃𝒍𝒆 (𝒚𝒆𝒕!)  Spearheaded by the U.S. Secretary of Commerce Gina Raimondo, AISIC represents a collaborative effort to mitigate the risks associated with AI, while capitalizing on its vast potential. The consortium's formation is a response to the urgent need for guidelines and standards in AI safety, focusing on areas such as red-teaming, risk management, and the watermarking of synthetic content. AISIC's establishment underscores the administration's commitment to harnessing AI's capabilities responsibly, ensuring that technological advancements do not outpace the frameworks designed to keep them in check. We see both sides, both the need for regulation, but also for it not to slow down this technology advancement. Curious how they'll continue to toe that line here. https://lnkd.in/gjB3BF8A

California's AI Safety and Innovation Bill https://lnkd.in/eWrxt4We With input from two most-cited AI researchers: Prof Geoffrey Hinton and Prof Yoshua Bengio, SB 1047 balances AI innovation with safety by: * Setting clear standards for developers of AI models with computing power greater than 1026 floating-point operations that cost over $100 million to train and would be substantially more powerful than any AI in existence today * Requiring developers of such large “frontier” AI models to take basic precautions, such as pre-deployment safety testing, red-teaming, cybersecurity, safeguards to prevent the misuse of dangerous capabilities, and post-deployment monitoring * Creating whistleblower protections for employees of frontier AI laboratories * Requiring transparent pricing and prohibiting price discrimination to protect consumers and to ensure startup developers have equal opportunities to compete and innovate * Empowering California’s Attorney General to take legal action in the event the developer of an extremely powerful AI model causes severe harm to Californians or if the developer’s negligence poses an imminent threat to public safety * Establishing a new public cloud computer cluster, CalCompute, to enable startups, researchers, and community groups to participate in the development of large-scale AI systems and align its benefits with the values and needs of California communities * Establishing an advisory council to support safe and secure open-source AI. It also mandates a "full shutdown" button before the training of an AI system starts.

Last week saw the creation of the US AI Safety Institute Consortium (AISIC) in support of the US AI Safety Institute at the US National Institute of Standards and Technology (NIST). According to NIST's press release, the AISIC "will unite AI creators and users, academics, government and industry researchers, and civil society organizations in support of the development and deployment of safe and trustworthy artificial intelligence (AI)." The AISIC comprises more than 200 participating companies to place the US as the leading country in setting AI safety standards and encouraging responsible AI innovation. While this is a step forward in supporting Executive Order 14110 on safe AI, there is an important balance to be struck between, on the one hand, over-regulating and, therefore, making it hard for all but the larger organizations to innovate in this space and, on the other hand, ensuring safe and responsible AI development. Applying Marc Andreessen's "simple plan" (https://lnkd.in/gqbJjYA7) may be a way to achieve this balance. #AIforGood https://lnkd.in/gvb2k7qa

AI governance is no longer just a buzzword—it's becoming a global priority. As artificial intelligence continues to advance at a rapid pace, the need for robust and comprehensive oversight becomes paramount. Australia's national AI regulation plan and the U.S. Commerce Department's proposed oversight are steps in the right direction. California's push for safety legislation is also noteworthy and could set a precedent for other states. The global focus on AI governance reflects an urgent necessity to ensure ethical standards and safety measures are in place. This isn't just about controlling technology; it's about fostering trust and responsibility in the AI industry. Read more about the latest developments in AI regulation and their potential impact on the industry: https://lnkd.in/e_j-E_EN

"For AI to be sustainable, it must be safe. As with any transformative technology, the risks imperil the benefits." California’s Safe and Secure Innovation for Frontier Artificial Intelligence Models Act, currently with Governor Newsom awaiting his signature, requires that developers of the most powerful AI systems write and follow safety and security protocols to prevent critical harm. According to one expert, Newsom should sign it. "Why California Gov. Gavin Newsom should sign an AI model regulation bill," by Anthony Aguirre, Faggin Presidential Professor for the Physics of Information at UC Santa Cruz and the executive director and co-founder of the Future of Life Institute. https://lnkd.in/eMiXfbcY

Californias new Artificial Intelligence Bill: “Test Test Test your AI models - functionally, safety, data privacy, ethically and free from bias and discrimination”! 🔥 Exciting news in the world of AI regulation! California's state Senate has passed a groundbreaking bill aimed at ensuring the safe development of large-scale AI systems. Senator Scott Wiener's SB 1047 establishes clear safety standards for developers of the most powerful AI models, focusing on those costing over $100 million to train.👏 The bill requires frontier AI developers to implement essential precautions, including pre-deployment safety testing, cybersecurity measures, and post-deployment monitoring. It also creates whistleblower protections and prohibits price discrimination to foster fair competition. Despite facing criticism from some tech industry voices, SB 1047 has garnered bipartisan support and backing from AI pioneers. As California leads the charge in balancing AI innovation with safety, this legislation could set a precedent for responsible AI development nationwide. With the bill now headed to the Assembly, it's clear that the Golden State is taking a proactive stance in addressing the potential risks and benefits of advanced AI technologies. 👌🏽 #AIRegulation #CaliforniaInnovation #TechPolicy #DataProtection #AIGovernance

Is it possible to test #LLMs to the point where regulators can deem them to be safe? The #UK's AI Safety Institute came out of the #aisummit2023. It's chaired by Ian Hogarth who has a great track record as an entrepreneur. This is refreshing for a government body. Ian is attracting some big names from the AI research community, and big tech itself. But the AISI is going to face three big problems. ⁉ Is AI genuinely testable? Logically, it's fundamentally different from conventional software as, in the case of AI, it's stochastic. It's hard for test teams to sign-off on deterministic systems, let alone algorithms that perform probabilistically. This raises questions about the nature of testing and what we mean by Safe. 📈 Scale and resources. If it takes tens-of-millions of dollars to train a single LLM how much does it take to test it? Will a government agency have access such resources? ⏳ Time. This appears to be what's bugging big tech. In theory, it would be great to have government bodies saying "your models are behaving well". However, there are now, what, 10+ serious contenders in the LLM arena? Each of these has a release and iteration cycle. What will the Service Level Agreement between AI Safety Institute and big tech look like? The AISI gave a good progress update, this week. https://lnkd.in/exFgDTHr The question is whether it can scale its activities to perform in line with big tech's schedule, because big tech is unlikely to wait. John Heaton-Armstrong, Ray Eitel-Porter, Emmanuel Ihenacho https://lnkd.in/edvFQnRF

The US has announced the #AI Safety Institute Consortium (AISIC). This initiative sees leading tech giants such as OpenAI, Google, Microsoft, Meta, Apple coming together. The AISIC is the fruit of Biden's executive order on enhancing AI practices across the board. This consortium is set to spearhead the development of comprehensive guidelines covering critical areas like red-teaming, capability evaluations, risk management, safety and security, and the watermarking of synthetic content. It's a clear signal that the future of AI is not just about technological breakthroughs but also about fostering trust and reliability in AI systems. #AI #Innovation #TechLeadership #AISafety #Collaboration

🌟 Along with more than 200 leading AI stakeholders, 𝐛𝐩 𝐰𝐢𝐥𝐥 𝐬𝐮𝐩𝐩𝐨𝐫𝐭 𝐭𝐡𝐞 𝐔𝐒 𝐀𝐈 𝐒𝐚𝐟𝐞𝐭𝐲 𝐈𝐧𝐬𝐭𝐢𝐭𝐮𝐭𝐞 𝐂𝐨𝐧𝐬𝐨𝐫𝐭𝐢𝐮𝐦 and its efforts to help advance the development and deployment of safe, trustworthy AI to create value for society and the economy. 🏆 Special shout out goes to our driving force of Responsible AI - Uthman Ali ! I am sure soon we will hear more exciting things in this space! https://lnkd.in/dscBV4Fh #bp #ai #responsibleai

📢 White House Announces Key Actions Following Landmark Executive Order on AI 🔥 Today the White House announced actions taken in the 90 days since President Biden issued the Executive Order on Safe, Secure and Trustworthy AI Key topics include: ➡ Managing Risks to Safety and Security -- Compel developers of powerful AI systems to report AI safety test results to the Department of Commerce -- Proposed reporting requirements for U.S. cloud companies that provide computing power for foreign AI training -- Completed risk assessments covering AI’s use in every critical infrastructure sector ➡ Innovating AI for Good -- Launched a pilot of the National AI Research Resource -- Launched an AI Talent Surge -- Began the EducateAI initiative -- Announced the funding of new NSF initiatives with a focus on advancing AI 📢 CAIDP Commentary: "We welcome the progress that the White House has made with the Executive Order on AI. These are important initiatives that deserve public support and the backing of Congress," said CAIDP President Merve Hickok. 👏🏽 👏🏽 "Still, we believe that more should be done to advance equity in AI initiatives across the federal government, and to ensure that the needs and rights of marginalized communities are not ignored." Ms. Hickok testified before the US Congress in March 2023 on the need for the United States to establish guardrails for AI and to develop expertise in government to manage the impact of AI on society. #aigovernance The White House #executiveorder https://lnkd.in/eAiJ88q9