Generative AI

A recent investigation highlights the business risks of using AI for web searches due to persistent data accuracy issues. While AI offers efficiency gains, a gap exists between user trust and technical precision, impacting compliance, legal defensibility, and financial forecasting. A study of six AI tools revealed accuracy varying from 55% to 71%, with all tools exhibiting errors, particularly in legal and financial advice. The lack of source transparency and potential for algorithmic bias further exacerbate risks. The report recommends companies implement governance frameworks, enforce prompt specificity, mandate source verification, and prioritize human oversight.

Radware (RDWR) launched its LLM Firewall, an addition to its Cloud Application Protection Services, on November 18, 2025. The LLM Firewall is designed to protect enterprises integrating Large Language Models (LLMs) by securing LLM prompts and responses against attacks like prompt injection and sensitive data exfiltration. It uses real-time, AI-based protection at the prompt level and a model-agnostic design to integrate across platforms and help comply with data protection standards. It addresses OWASP Top 10 Risks for LLMs and Gen AI Apps.

This Monday’s focus is on Salesforce (CRM) and Goldman Sachs (GS). Analysts have mixed views on Salesforce’s future due to concerns about AI’s impact, despite reaffirming “Buy” ratings. CRM’s stock is down 29% YTD, highlighting investor uncertainty. Goldman Sachs is dominating M&A, capturing a 34% market share and advising on significant deals. GS is on track for its best M&A performance in 25 years, with its stock up nearly 35% YTD. The key is if GS can maintain dominance amidst evolving economic conditions.

SoftBank sold its entire Nvidia stake for $5.83 billion in October, not necessarily due to valuation concerns. Instead, the move signals a strategic shift, with proceeds reinvested into generative AI, specifically OpenAI. This reallocation underscores SoftBank’s confidence in generative AI’s long-term potential over a diverse tech portfolio. The decision highlights a concentrated approach, prioritizing transformative AI solutions and reflecting their conviction that substantial investment is needed to unlock these technologies.

SoftBank divested its Nvidia stake for $5.83 billion to reallocate capital to OpenAI, signaling a firm belief in generative AI’s potential. Despite market discussions on AI valuations and concerns raised by figures like Michael Burry, SoftBank’s decision reflects a strategic prioritization of investment, aiming to capitalize on perceived greater opportunities within OpenAI. This move underscores SoftBank’s conviction that OpenAI is poised for significant growth across various sectors, making the increased investment a strategic imperative.

Microsoft is forming the MAI Superintelligence Team, led by Mustafa Suleyman, to research advanced AI with a focus on practical, controllable applications designed to serve humanity. This initiative, backed by significant investment, aims to develop AI companions for education, medicine, and renewable energy, contrasting with the pursuit of generalist AI. The team will diversify AI sources beyond OpenAI and prioritize a “humanist” approach to superintelligence, emphasizing ethical considerations and clear boundaries to ensure responsible and beneficial advancements.

A new architectural design, Continuous Autoregressive Language Models (CALM), offers potential cost savings for enterprises deploying AI. CALM predicts continuous vectors instead of discrete tokens, compressing information and reducing computational steps. Experiments show CALM models achieve comparable performance to baselines with significantly fewer FLOPs. This novel approach requires a new “likelihood-free framework” including training methods, a BrierLM evaluation metric, and a likelihood-free sampling algorithm. CALM highlights a shift towards architectural efficiency as a crucial factor in reducing enterprise AI costs and improving sustainability.

NYU Stern MBA students are optimistic about AI’s impact, seeing it as a tool rather than a threat, according to Professor Robert Seamans. His curriculum emphasizes hands-on AI experimentation, including exercises where students use LLMs to build and dismantle arguments, fostering critical thinking. While high school and college students are more cautious, viewing AI as an aid, not a replacement for core skills, Georgia Tech is adapting curricula to emphasize higher-level concepts that AI cannot easily solve. Successful organizations will cultivate continuous learning and leverage AI to augment human ingenuity.

OpenAI’s AI video generation app, Sora, has launched on Android in the U.S., Canada, Japan, South Korea, Taiwan, Thailand, and Vietnam, following its iOS debut. The Android expansion targets a larger global audience, aiming to drive adoption and refine AI algorithms. Sora allows users to create videos from text prompts, fostering a creative community similar to TikTok. Analysts predict a significant impact on creative fields due to Sora’s accessible and advanced generative AI video capabilities.

Wall Street saw a volatile Wednesday following the Fed’s rate cut and Chairman Powell’s cautious remarks. The S&P 500 briefly hit an intraday high, driven by semiconductor strength fueled by AI demand. Nvidia’s market cap surpassed $5 trillion. Analysts are revising upwards, anticipating sustained AI capital expenditure. Boeing shares declined on 777X delays. Honeywell’s Solstice spin-off finalized, while analysts are positive on DuPont’s Qnity spin-off. Investors await earnings from Meta, Microsoft, and other major companies.