🌍 Anthropic's Deloitte Deal, Enterprise AI Success Rates, and Your Chapter Events This Week

Inside: New Dallas Chapter Co-President, AI liability frameworks, and why 95% of AI pilots fail

Author

Applied AI Association
October 06, 2025 • Estimated Reading Time: 23 minutes

👋 Editor's Note

This week’s AI Beat newsletter brings significant enterprise AI news, starting with Anthropic's largest deployment to date, rolling out Claude to Deloitte's 470,000+ employees worldwide. We're also diving into why 95% of enterprise AI pilots fail and what the successful 5% do differently, plus examining the real costs of AI implementation as organizations grapple with budget overruns exceeding 50%.

We're excited to spotlight Prakash P. Menon as our new AAIA Chapter Co-President for Dallas, bringing deep expertise in energy sector digital transformation. AAIA India member Subathra Mylsamy tackles a critical question in her article: "Who Owns the Mistake?" when AI-driven decisions cause harm.

Your local chapters are active this month with events ranging from AI strategy in healthcare (India Chapter, Oct 7) to job search optimization with AI (Portland/SW Washington Chapter, Oct 15). We've also included important updates on AI regulation, from California's new safety disclosure law to the NHS's accelerated AI adoption framework. Check out our full calendar of upcoming events, recent recordings, and industry insights below.

Have something to share with the global AAIA community? Send your news, tools, or events to [email protected].

📅 Upcoming events:

🗓️ India AAIA Chapter

🗓️ Portland/SW Washington AAIA Chapter

🗓️ Portland/SW Washington AAIA Chapter

🗓️ VR/AR Association and Applied AI Association

🗓️ Charlotte AAIA Chapter

🗓️ Seattle AAIA Chapter

  • 🗓️ November 10 - 14, 2025

  • Google is offering a free, five-day online AI Agents Intensive course from November 10–14, 2025, which includes daily assignments, live seminars, and a capstone project to help participants build and deploy AI agents. (show me more)

  • 🔗 Learn more and register

🗓️ AAIA partner TechCon Global’s events, focused on the latest trends, advancements, and challenges across the technology landscape, including Life Sciences, Digital Health, Artificial Intelligence, Robotics, and Semiconductors:

📢 AAIA chapter and member news and updates:

🌟 AAIA Chapter Co-President Spotlight: AAIA is pleased to announce that Prakash P. Menon has been appointed as AAIA Chapter Co-President for Dallas, TX, USA.

About Prakash:

Prakash P. Menon is a seasoned technology and industry leader with deep expertise in the Energy & Utilities sector, specializing in digital transformation, grid modernization, smart water, clean gas and AI-driven innovation. He has a strong track record of aligning technology strategy with business outcomes to drive operational efficiency, resilience, and customer-centric growth.

💬 Reflecting on his new role, Prakash said:

"I’m excited to serve as AAIA Chapter Co-President because it’s an opportunity to bring people together, spark innovation, and create lasting impact. This role allows me to channel my passion for technology and leadership into building a stronger community, inspiring the next generation, and shaping the future of our industry.

I have seen the transformative impact of innovation, collaboration, and shared learning, and this role allows me to foster that same spirit within the AAIA network. By bringing together diverse professionals, driving meaningful discussions, and creating platforms for knowledge exchange, I look forward to helping our chapter become a hub of thought leadership and impact. Most importantly, it is a chance to give back, inspire the next generation of leaders, and help shape the future of our industry together.”

Welcome aboard, Prakash!

📰 AAIA Member article, by Subathra Mylsamy, Managing Partner, AK Mylsamy & Associates LLP

[Article excerpt; read the article in its entirety on the AAIA website.]

Who Owns the Mistake? Rethinking Liability in AI-Driven Decision-Making

by Subathra Mylsamy, Managing Partner, AK Mylsamy & Associates LLP

A problem of attribution and consequences

We live in a moment when machine decisions increasingly substitute for or augment human judgment. An AI model embedded in a vehicle can decide whether to brake or swerve in milliseconds; a diagnostic algorithm can suggest cancer treatments; an automated trading system can execute thousands of orders in a second. When those decisions cause harm, the central legal question becomes deceptively simple: who owns the mistake? The answer matters, not only for victims seeking compensation, but for social incentives that shape how firms test, deploy, and update AI systems. Scholars, regulators and courts have begun to confront this question, but a patchwork approach remains the norm: agencies issue sectoral guidance, courts stretch old tort categories to new technology, and legislators debate whether bespoke AI liability rules are needed.

The doctrinal starting points: product liability and professional negligence

Product liability. A natural starting point is product liability. This body of law, familiar from cases involving defective machines or cars, traditionally imposes responsibility on manufacturers for three types of defects: manufacturing, design, and failure to warn. In many situations, manufacturers can be held strictly liable, meaning victims do not have to prove negligence. At first glance, applying this framework to AI makes sense—especially where AI is embedded in physical products such as cars or medical devices. Yet AI differs in fundamental ways. Unlike static products, AI systems change over time through updates and continued learning, and they depend on outside data and cloud services. The familiar tools of product liability law, designed for fixed objects, struggle to keep up with these shifting systems. In the United States product liability traditionally addresses harm caused by defective products. Beginning with landmark judicial decisions (e.g., Greenman v. Yuba Power Products), and refined in the Restatement (Third) of Torts: Products Liability, the doctrine recognizes three defect categories: manufacturing defects, design defects, and warning (failure-to-warn) defects. Liability can be strict (no need to prove negligence) for manufacturing defects and, depending on the jurisdiction and test applied, for design defects when a product is unreasonably dangerous. These principles place the cost of accidents on manufacturers that put dangerous goods into the stream of commerce.

Applied to AI, product liability’s appeal is obvious: many AIs are embedded in physical devices (cars, medical devices) and cause compensable physical injury. But traditional product tests were built for physical objects with static designs; AI systems are often data-driven, opaque, may change through updates, and depend on third-party data and cloud services. These characteristics complicate applying standard defect analyses and causation tests…

Read the rest of the article at:  https://members.aaiaglobal.com/news/Details/aaia-india-member-article-who-owns-the-mistake-rethinking-liability-in-ai-driven-decision-making-292256

📰 In the News:

The article explains why 95% of enterprise AI pilots fail due to misaligned expectations, fragmented efforts, and lack of human-centered strategies, and it introduces the TRAILS framework—a human-first roadmap emphasizing trust, clear purpose, rapid prototyping, cultural mindset, continuous learning, and scaling—which successful organizations use to achieve lasting AI transformation and surpass common pitfalls. (show me more)

Anthropic announced it will triple its international workforce and expand its applied AI team fivefold in 2025 to meet surging global demand for its Claude AI models, which now see nearly 80% of usage outside the U.S., as the company accelerates its worldwide enterprise expansion and deepens competition with industry leaders. (show me more)

Anthropic has launched Claude 4.5, a new AI model designed for business and enterprise customers, featuring significant advances in autonomous, uninterrupted coding (up to 30 hours), superior performance on financial and scientific reasoning tasks, and a targeted focus on practical, extended use cases over viral consumer appeal, positioning it as a leading enterprise AI solution. (show me more)

The article highlights that AI project cost overruns are increasingly common, with many organizations severely underestimating expenses, often by more than 50%, which causes budget shortfalls, delays, and reputational risks for CIOs, making cost transparency and strategic FinOps practices essential for effective AI implementation and sustained stakeholder trust. (show me more)

JPMorgan Chase is transforming into the world’s first fully AI-powered megabank by integrating advanced large language models and AI agents into every aspect of its operations, aiming to enhance employee productivity, automate complex tasks, reduce costs, and revolutionize client interactions, though full implementation and workforce impacts will take several years to materialize. (show me more)

Walmart CEO Doug McMillon states that AI will fundamentally change every job within the company, automating some tasks while creating new roles, and plans to keep its workforce size stable by focusing on employee upskilling and adapting job functions to leverage AI rather than replace people entirely. (show me more)

Lufthansa announced plans to cut 4,000 administrative jobs by 2030, primarily in Germany, as part of a broad digital transformation strategy that leverages artificial intelligence and automation to boost efficiency, reduce costs, and enhance profitability while simultaneously investing in fleet renewal and digital innovation across its airline group. (show me more)

Deloitte has announced its largest enterprise AI deployment to date by integrating Anthropic's AI assistant Claude across more than 470,000 employees worldwide, establishing a Claude Center of Excellence for training and support, and developing industry-specific AI solutions focused on compliance and productivity enhancement in regulated sectors. This expanded partnership underscores Deloitte's commitment to responsible, scalable AI adoption and positions the firm to better advise clients on AI transformation across industries globally. (show me more)

📰 Commission launched to help accelerate NHS use of AI (DigitalHealth, September 26, 2025)

A new UK national commission, including experts from big tech, clinicians, and regulators, has been launched to create a regulatory framework aimed at accelerating safe and responsible adoption of AI technologies in the NHS, with a regulatory rulebook planned for 2026 to support innovation and patient safety in healthcare AI. (show me more)

The article summarizes how states, notably Illinois, are establishing a patchwork of inconsistent laws—including Illinois' outright ban on using AI for mental health treatment—to prioritize public safety and ensure licensed professionals deliver therapy, even as critics warn these rapid regulations may stifle beneficial AI innovation and fail to govern general, unregulated chatbots. (show me more)

The article explores the transformative role AI professors will play in higher education, emphasizing how AI-driven avatars will provide personalized, accessible, and evidence-based teaching that complements human professors by enhancing interdisciplinary learning, critical thinking, and student engagement within a hybrid educational model supervised by humans to mitigate AI limitations. (show me more)

Meta will begin using conversations with its AI chatbot across Facebook, Instagram, Messenger, and Ray-Ban smart glasses to personalize ads and recommendations starting December 16, 2025, without an opt-out option for most users, raising significant privacy concerns about the exploitation of private AI chats for targeted advertising. This change formalizes Meta's existing personalization strategy but blurs privacy boundaries by mining AI chat data, though sensitive topics will not be used for ads, and users will be notified starting October 7. (show me more)

📰 CAMIA privacy attack reveals what AI models memorise (AINews, September 26, 2025)

The CAMIA privacy attack is a novel, context-aware method that effectively detects sensitive data memorization in generative AI models by analyzing their token-level prediction uncertainty, revealing significant privacy risks in AI training and urging the development of better privacy-preserving techniques. (show me more)

California Governor Newsom signed SB 53, a law requiring large AI companies to publicly disclose their plans for mitigating catastrophic risks (like runaway models or bioweapon aid), imposing fines of up to $1 million per violation and aiming to fill the regulatory gap left by Congress. (show me more)

📰 This Week’s Slighty-off-topic Reads:

The article discusses the controversial debut of Tilly Norwood, an AI-generated actress created by Particle6, highlighting the mixed reactions within Hollywood including backlash from actors and unions, interest from some talent agents, and the broader debate on AI's role in entertainment as both a creative tool and a potential threat to human performers. (show me more)

A Chinese robotics company, AheadForm, has developed an ultra-lifelike robot head called Origin M1 that can blink, twitch, nod, and exhibit subtle human facial expressions using 25 micro motors and integrated sensors, aiming to improve human-robot interactions in sectors like customer service and healthcare, while raising ethical and psychological debates about lifelike machines. (show me more)

Watch the demo video: https://www.youtube.com/watch?v=w4kC-XCEXaQ

🛠️ 📘 Tools, Charts, and Reports Spotlight:

The report analyzes the rapid and uneven adoption of AI technology worldwide, highlighting its concentration in high-income regions and specialized tasks such as coding, with enterprise use focused heavily on automation via APIs; it emphasizes that while AI adoption is accelerating productivity and transforming economic activity, significant geographic and sectoral disparities raise concerns about widening inequalities and the need for strategic policymaking to ensure broad-based benefits from AI-driven economic growth. (show me more)

📘 What Companies with Successful AI Pilots Do Differently (Harvard Business Review, September 12, 2025)

The article explains that the key factor distinguishing the 5% of companies with successful AI pilots from the 95% that fail is effective leadership, specifically, leaders who act as "AI shapers" by embedding AI into business strategy, building trust, driving adoption, and demonstrating five key behaviors (strategic agility, human centricity, applied curiosity, performance drive, and ethical stewardship), and it offers a framework and practical guidance for developing these leadership capabilities to scale AI transformation beyond pilots. (show me more)

If you have an event, newsworthy story, industry insight/report, or AI tool to share with AAIA members, please email the details to [email protected].

Our list of AAIA Chapters is continually expanding—if you live in a Chapter city or region and want to get involved, please email that city/region’s Chapter President.

Become an AAIA member and join the leading global industry association dedicated to integrating AI technologies into business. Accelerate knowledge exchange, collaboration, and innovation through our worldwide network of chapters, specialized industry committees, and in-depth reports. Connect with thought leaders and innovators to help shape the future of AI in business. Visit our website to learn more about membership opportunities.

Questions?  Email us at [email protected].

Until next time!