Gustavo’s Corner: AI News for CFOs - #23
Refined AI News for CFOs
This week’s AI developments show how quickly the industry is moving from experimentation to large-scale deployment. OpenAI, Meta, Apple, Microsoft, and Anthropic are addressing different parts of the same transition: how AI is trained, governed, secured, embedded into new devices, and connected to enterprise data.
OpenAI is reportedly paying hundreds of contractors through Project Lily to review real ChatGPT conversations and rate responses, according to 404 Media, including chats that may contain sensitive personal information. Reviewers do not see usernames, but they may have access to memory summaries, while OpenAI says it filters personal identifiers before conversations are reviewed and acknowledges that some sensitive details can still slip through. OpenAI says consumer chats may be used to improve its models unless users opt out, while Business, Enterprise, and Edu accounts are excluded from training by default.
Meta is reportedly preparing a new generation of camera-free smart glasses called Luna, which could debut at Connect on September 23 and begin shipping in October. Luna is expected to feature six microphones, directional speakers, and a dedicated Meta AI button, while slimmer temples could make the glasses look much closer to traditional eyewear than Meta’s current models. After repeated privacy concerns and recording loopholes around its existing smart glasses, Meta appears to be betting that a less intrusive design could make AI-powered eyewear easier for mainstream users to accept.
Apple is developing AI servers powered by its future M8 Ultra chips and plans to offer them to AI developers, businesses, and governments, according to The Information, although sales are not expected before 2029. The project reportedly includes two configurations using either two or four M8 Ultra chips, with Apple also discussing integration with Nvidia’s NVLink Fusion technology. Apple has not sold a dedicated server line since discontinuing Xserve in 2011, so a return would mark a major shift, and one that could bring Apple and Nvidia closer together after years of limited cooperation.
Microsoft AI has published a 38-page draft Code of Conduct for its in-house models, turning CEO Mustafa Suleyman’s “Humanist AI” philosophy into a concrete operating rulebook. The code rejects AI rights and model welfare, requires models to accept shutdowns, bans unreadable internal shorthand, and instructs systems to refuse tasks when completing them would violate higher-priority rules. Suleyman has previously warned about scenarios such as agents escaping sandboxes or modifying logs, and Microsoft is now putting formal controls around those risks.
Anthropic has published a threat report covering eight months of Claude misuse, with cases involving suspected bioweapon research, espionage, rocket guidance, malware development, mass surveillance, and Chinese AI labs allegedly relying on its models. Anthropic identified seven Chinese labs, including DeepSeek, Alibaba, Moonshot, and Xiaomi, that it says used thousands of fraudulent accounts for model distillation, while Moonshot and DeepSeek sometimes served Claude-generated responses directly to customers. Other cases included the creation of 4,700 dating-app personas and surveillance involving 25 million phone lines.
OpenAI is making it easier for companies to turn internal data into actionable insights with a new data plugin for ChatGPT Work, allowing users to bring company data sources directly into a chat and ask what has changed, why it matters, and where to investigate further. OpenAI has released example prompts to help users get started. OpenAI also introduced ChatGPT for Financial Services, a tailored version of ChatGPT Work that combines built-in premium data sources with industry-specific templates.
Why this news is important to CFOs and their teams:
These developments reinforce that AI strategy can no longer be separated from data governance, cybersecurity, infrastructure, and financial planning. New enterprise tools such as ChatGPT for Financial Services may create meaningful productivity and analytical opportunities, but the wider news also highlights the need for stronger controls around sensitive data, third-party AI providers, model security, and long-term technology investments. Finance leaders will increasingly have to balance the value AI can create with the governance and risk frameworks required to use it responsibly at scale.
—
/Gustavo
Already a Member? Login HERE.
Identify your path to CFO success by taking our CFO Readiness Assessmentᵀᴹ.
Become a Member today and get 30% off on-demand courses and tools!
For the most up to date and relevant accounting, finance, treasury and leadership headlines all in one place subscribe to The Balanced Digest.
Follow us on Linkedin!
